TeleGeography 1996/97 Global Telecommunications Traffic Statistics and Commentary This work is based upon sources believed to be reliable, but the pub- lisher does not warrant the accuracy or completeness of any informa- tion for any purpose and is not responsible for any errors or omissions. This work is for the confidential use of subscribers. No part may be reproduced, stored in an information retrieval system, or transmitted, in any form or by any means, electronic, mechanical, photocopied, record- ed or otherwise, without written consent from the publisher. All rights reserved. Copyright © TeleGeography, Inc. 1996 First Printing--October 1996 Second Printing--December 1996 Third Printing--January 1997 This is copy Do not reproduce. ISBN 1-886142-08M4 Printed in the United States of America TeleGeography, lnc ¯ Suite 1000 I 150 Connecticut Avenue, NW ¯ Washington, DC 20036 USA Cover Illustration: Annemarie Feld and Gregory C. Staple © TeleGeography, Inc. 1996 All rights reserved. telegeography \t~l’a-j~-0g’r~-fO\ n (1990) abbrv, of telecommunications geography [fr. Gk tele, far off, at a dis- tance and L. communicatus, pp. of communicare to impart ÷ fr. Gk geo (earth) ÷ graphein, (tO write)] 1. a new branch of geography that maps the pattern of telephone traffic and other electronic communication flows; 2. places created by or perceived solely via telecommunications (e.g., a computer network address); 3. the telecommunications artifacts (radio antennae, terminals, signs) on a site; 4. the balance of telecommunications power in one country or region vis-h-vis another (cf. geopolitics, archaic). About the Cover The Egyptian astronomer Ptolemy cataloged about 1,022 stars, grouped into 48 constella- tions, during the 2nd century C.E. Another 40 constellations were recognized beginning in the 16th century, as European navigators widened their explorations. Coelum (the chisel), Horologium (the clock), and Octants (the Octant, for John Hadley’s 1730 invention), joined Orion; Ursa Major; and Leo in the heavens. Times change. And, with them, the way we picture the night sky. Send us your favorite "dig- ital" constellations and we’ll send you a complimentary TGI publication. © TeleGeography, Inc. 1996 TeleGeography t996/97 TeleGeography, Inc. This report was prepared by TeleGeography, Inc., an independent research and publishing company, in cooperation with the International Institute of Communications (IIC), a London-based non- profit educational and conference organization. The TeleGeography 1996/97 production team included: Gregory C. Staple ..................... Editor Jason Kowal ................... Deputy Editor Zachary M. Schrag ........... Technical Consultant Barbara Frank ........... Administrative Assistant Annemarie Feld ............... Design Consultant Research assistance and fact checking were provided by: Stephan Beckert, Barbara Dooley, Nancy Dysart, Eva Pregon and lan Watson. The following also are available from TeleGeography, Inc.: New International Carriers 1996 (ISBN 886142-05-X) The TeleGeography 1 O0 (ISBN 1-886142-06-8) Direction of Traffic 1996 (ISBN 92-61-06291 1 ) Telecommunications Map of the World (ISBN 1-86186-005-6) For a current publications list contact: TeleGeography, Inc. Suite 1000 1150 Connecticut Avenue, NW Washington, DC 20036 USA Tel.: +1 (202) 467-0017 Fax: 1 (202) 467-5915 E-mail: info@telegeography.com Or visit our web site at: http://www.telegeography.com For membership information and details on IIC publications, contact: The International Institute of Communications Tavistock House South Tavistock Square London WC 1H 9LF, England, UK Tel.: +44 171 388 0671 Fax: +44 171 380 0623 e-mail: iic@clusl ,ulcc.ac.uk III TeleGeography 1996/97 © TeleCeography, Inc. 1996 © TeleGeography, Inc. 1996 TeleGeography 1996/97 Contentsl Preface ............................................................. ix Acknowledgments ...................................................... x Introduction .......................................................... xi Beyond the Public Switched Network Graham Finnie and Melanie Stockwell Gregory C. Staple The Editors Sizing the Market for International Private Line Service ....... 3 A Primer on International Simple Resale (ISR) ............ 15 International Accounting Rates ....................... 22 Settlements for International Phone Sex ................ 24 The lnternet The Editors K. Cla~ Zachary M. Schrag Robert Shaw The Editors Internet Traffic Statistics: An Overview ................. 27 Closing the lnternet Statistics Gap .................... 31 The Achilles Heel of Internet Telephony ................. 37 Internet Domain Names ........................... 41 Mapping the Internet ............................. 49 International Facilities and Carriers Map of Major Submarine Cables ........................................... 60 Proposed Fiber Optic Cable Systems ........................................ 62 Intercontinental Cable and Satellite Capacity (1986-2000) ......................... 62 The Next Generation of Cables: Beyond 50 Gbps ............................... 63 The Top 40 International Carriers ........................................... 64 The Top 50 International Routes ........................................... 65 Market Share of Competing International Carriers ............................... 66 International Traffic Statistics Outgoing International Traffic, by country Americas ..................................... 71 Europe ........................................ 91 Asia, Middle East and Africa ....................... 123 Methodology and Sources .............................................. 154 Blue Pages International Telephone Dialing Codes ....... A~3 North American Area Codes ............. A-6 World Dialing Codes Map ............... A-9 World Time Zones ................... A-1 1 Political Geography .................. A-1 3 National Telecommunications Indicators ..... A-14 Gibson’s Geographica .............................................. End Page TeleGeography 1996/97 © TeleGeography, Inc. 1996 Traffic Tables Americas Argentina ...................... 73 Bolivia ....................... 74 Brazil ........................ 75 Canada ....................... 76 Chile ......................... 77 Colombia ...................... 78 Dominican Republic .............. 79 Ecuador ...................... 80 Mexico ....................... 81 Paraguay ...................... 82 Peru ......................... 83 United States, main outgoing routes .84 United States, main incoming routes .85 United States, other correspondents ...86 United States, traffic by carrier ....... 87 Uruguay ...................... 88 Venezuela ..................... 89 Europe Austria ....................... 92 Belgium ....................... 93 Croatia ....................... 94 Cyprus ....................... 95 Czech Republic .................. 96 Denmark ...................... 97 Estonia ....................... 98 Finland ....................... 99 France ....................... 100 Germany ..................... 101 Greece ...................... 102 Hungary ..................... 103 Iceland ...................... 104 Ireland ...................... 105 Italy ........................ 06 Luxembourg ................... 07 Macedonia, TFYR ............... 08 Netherlands ................... 109 Norway ...................... 1 0 Poland ...................... 11 Portugal ..................... 112 Russia ....................... 1 3 Slovak Republic ................ 114 Slovenia ..................... 1 1 5 Spain ....................... 1 1 6 Sweden ...................... 117 Switzerland ................... 118 Turkey ....................... 119 United Kingdom ................ 120 Yugoslavia .................... 121 Traffic within the Former Soviet Union .................. 122 Asia, Middle East, and Africa Australia ..................... 125 Bahrain ...................... 26 Bangladesh ................... 127 China ....................... 128 Hong Kong .................... 129 India ........................ 130 Indonesia ..................... 131 tran ........................ 132 Israel ....................... 133 Japan ....................... 134 Jordan ...................... 135 Korea, Republic of .............. 136 Kuwait ....................... 137 Macau ...................... 38 Malaysia ..................... 139 Morocco ..................... 140 New Zealand .................. 141 Pakistan ..................... 142 Philippines .................... 143 Qatar ....................... 144 Saudi Arabia .................. 145 Singapore .................... 146 South Africa ................... 147 Sri Lanka ..................... 148 Syria ........................ 49 Taiwan ....................... 150 Thailand ..................... 151 United Arab Emirates ............ 152 Vietnam ...................... 153 © TeleGeography, Inc. 1996 Te]eGeography 1996/97 Other Tables and Figures Introduction International Traffic Continues to Grow Faster than the Number of Lines xi Trends in the Global Information Economy xii Eat or Be Eaten: North American Carriers Are at the Top of the Globa! Food Chain xiii How Do U.S. Carriers Use Their International Facilities? xiv The Top 20 Carriers: Traffic Growth xv Richer Countries Call Out, Poorer Countries Have a Traffic Surplus xvii Beyond the Public Switched Network Sizing the Market for International Private Lines 1995 International Telecommunications Revenue How Much Do Leased Circuits Cost? Five Percent of the World’s Telephone Traffic is Carried by Private Networks Top Ten IPL Routes Corporate Demand for International Services (1996) Estimated IPLs, by Country Major IPL Suppliers Oftel’s Cost Equation The World’s Largest Private Networks The Big Network: SITA’s 1200-Node Monster Westinghouse: A Middle Sized Corporate Network International Simple Resale Accounting Rates and How They Work Sample Break-Even Calculation for ISR Carrier Who Cares Most About ISR? Legal Rules for ISR in Selected Countries The Impact of ISR: Carriers, Traffic and Accounting Rates How Do U.S. Carriers Use Their International Facilities? Call-back -- Cash-back: Why Call-back May Help Low Income Countries International Accounting Rates United Kingdom Accounting Rates for OECD Countries, 1991-1996 Accounting Rate Trends for OECD Countries, 1991-1996 United States Accounting Rates, 1992-1996 Major Providers of International Audiotex Global Audiotex Traffic Growth, 1992-1996 Settlement Rates for Audiotex Providers 3 4 5 6 7 7 8 9 10 12 13 16 17 17 18 19 19 20 22 22 23 24 24 24 The lnternet lnternet Traffic Statistics: An Overview The Traffic Statistics Dry Up Mr. Quarterman’s Internet Weather Report lnternet Host Growth, 1981-1996 Closing the Internet Statistics Gap The NSF Transition Internet Metrics and Tools The Internet Engineering Task Force World Wide Web Traffic as a Percentage of Total Internet Packets 27 28 29 33 34 38 VII TeleGeography 1996/97 f IOther Tables and Figures (continued) © TeleGeography, Inc. 1996 The Internet (continued) Internet Domain Names The Internet’s Unelected Governors Domain Names and Trademarks International or National Top Level Domains? Mapping the Internet Internet Cash Flows lnternet Packet Flows: End Users to Global Backbones lnternet Exchanges: Where the Backbones Meet The Internet’s Big Switches: Major U.S. Network Access Points (NAPs) High-Speed lnternet Backbones, 1995-96 North American lnternet Backbones European Internet Backbones Asian Internet Backbones Traffic to and from the U.S. via FIX-West, July 1996 42 43 44 45 49 5O 51 52 53 54 55 55 56 Facilities and Carriers Major Submarine Cables Host Computer Growth by Domain, 1995-1996 Proposed Fiber Optic Cable Systems Cable and Satellite Capacity on Trans-Atlantic and Trans-Pacific Routes, 1986-2000 The Next Generation of Cables: 100 Gbps and Beyond The Top 40 International Carriers, 1990-1995 The Top 50 International Routes, 1995 Market Share of Competing International Carriers, 1988-1995 6O 57 62 62 63 64 65 66 VIII © TeleGeographyo Inc. 1996 TeleGeography 1996/97 Preface s the dramatic transformation of the global telecommuni- cations industry continues, a new wholesale market for trans-border services has emerged and is growing rapidly. Liberalization is releasing the regulatory brakes on the of international services, and the global are developing their offerings to address this market, both within and outside of the traditional trading methods of the correspondent regime. This new market is diverse, including the ever growing number of re- sellers and call-back operators, emerging "Second Operators" and the large numbers of as yet unaligned Telecommunications Operators [TOs) looking for lower cost international delivery. Then there are of course the behemoth "domestic players" such as the U.S. Regional Bell Operating Companies [RBOCs), who have yet to make clear their impact upon the international scene. As the pace of change in the telecoms and related industries contin- ues to increase, the importance of reliable information comes ever to the fore. TeleGeography has, since its launch in 1989, been hailed as the basic industry reference text on international traffic flows. It is a reliable information source that is essential for informed decision mak- ing by companies at all levels in the market, from those forming glob- al alliances to those exploiting short term arbitrage opportunities as they arise around the globe. In addition, this edition provides invaluable insights into the evolution of this market and the impact of new technologies and alliance devel- opments, and it particularly focuses upon the development and impli- cations of the lnternet. TeleGeography continues to be an invaluable industry tool which MCI and BT are proud to co-sponsor. Timothy E Price President and Chief Operating Officer MCI Telecommunications Corporation Alfred T. Mockett Managing Director BT Global Communications TeteGeography 1996/97 Acknowledgments © TeleGeography, Inc. 1996 W ~e are indebted to the numerous carriers, government departments, regulators and international organizations from around the world who responded to our requests for information. Special thanks are due particularly to the dedicated people at these organizations and at the International Telecommunication Union (ITUI in Geneva who set aside the time necessary to ensure that the data reported here are as current and accurate as possible. This year’s edition of TeleGeography was also supported, in part, by publication grants from MCI Communications Corporation and British Telecommunications plc. As in the past, however, these grants were made without any precondition; TeleGeography, Inc. is solely respon- sible for the report’s editorial contents. The Editors © TeleGeography, Inc. 1996 Introduction Introduction "People prefer that which is complicated, growing and sufficiently unpredictable to be interesting." E.O. Wilson, Biophilia (1984) A s an ecosystem, the global communications busi- ness has few peers either in size or complexity. Its networks span the earth offering services as different as facsimile and television, e-mail and imobile telephony. Electronic communication networks now touch our lives whether at work or at play, in our home or in our car. As in other ecosystems, the major subsystems are intimately related though the linkages may not be readily apparent. This edition of TeleGeography pro- files two of these symbiotic worlds: the public telephone net- work and the world of international private lines. The global telephone network is primarily used for voice communications. Private networks, by contrast, are mainly used for data communications. The Internet, for example, is the world’s largest private data network. However, these lnternational Traffic Continues to Grow m I000 9O0 L Fixed main lines 8OO ¯ Total international traffic 7OO two networks are marked by a type of reciprocal evolution- ary change or co-evolution which affects us all. The price of international calls; the success of telecommunication trade talks; the future for Internet telephony; and access to multi- media services--these are but a few of the issues shaped by how the world’s pdvate and public networks interact. The seeds of today’s co-evolutionary cycle were planted in the early 198Os when regulators in the U.S. and several other countries began to free private data networks from the rules governing public telephony. As the regulatory environment for these two networks diverged, both net- works entered a period of unprecedented growth even as they became more specialized. International telephone traffic grew fivefold from 1985 to 1995 (see Figure 1 ) and the average price for overseas calls Faster than the Number of Lines 600 500 projected (12% growth) 84.7 94.9 106.3 lOO 6O == o o 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000~ Note: Data include outbound Minutes of Telecommunication Traffic (MITT) on public telephone networks only. Source: ITU, TeleGeography, Inc. ©TeleGeography, Inc. 1996 Xl TeteGeography 1996/97 © TeleGeography, Inc. 1996 Figure 2. Trends in the Global Information Economy Traffic growth trends, 1986-1995 and projections, 1995-2000 Historical trend 10% growth 12% growth 14% growth CAGR CAGR CAGR CAGR Indicator 1986 1995 1986-95 2000 1995-2000 2000 1985-2000 2000 1995-2000 Calls (Bn) 3.6 16.8 18.5% 33 14.5% 35.8 16.3% 39.1 18.4% Estimated call length (rains) 4.6 3.6 -2.7% 3 -3.5% 3 -3.5% 3 -3.5% Minutes (Bn) 16.7 60.3 15.4% 98.9 10.4% 107.3 12.2% 117.2 14.2% Per main line subscriber 38.8 87.4 9.4% 109.9 4.7% 116 5.8% 123.4 7.1% Per main line plus mobile 38.6 77.8 8.1% 82.4 1.2% 85.8 2.0% 90.1 3.0% Revenue 21.7 55 10.9% 73.6 6.0% 75.5 6.5% 78.3 7.3% Assumptions Price per MiTE ($) 1.3 0.91 -3.9% 0.74 -4.0% 0.7 -5.0% 0.67 -6.0% Main lines (M) 430 690 5.4% 900 5.4% 925 6.0% 950 6.6% Mobile subscribers (M) 1.4 85 58.1% 300 28.7% 325 30.8% 350 32.7% Note: Source: 1986-1995 based on reported data. 1995-2000 based on ITU forecasts, Scenarios are as follows 1. 10% growth: Contznuing trend of lastfour years. 2. 12% growth: Assuming a faster network growth rate and faster rates of price-cutting. 3 14% growth. Assumption that recent slowdown w~ll be decisively reversed towards the end of the decade due to a combination of price cutting, demand stimulation and the development of new services A sigmficant component of this new demand would be traffic generated from mobzlem subscribers. Direction of Traffic 1996 (ITU/TeleGeography, Inc.) © ITU/TGI 1996 fell to $0.75 per minute or less on many routes. Data net- working by multinational companies grew even more rapid- ly. By the early 1990s, a vast consumer market also devel- oped for the lnternet as new protocols allowed multimedia information to be transmitted with other data on private computer networks. networks to trial next generation technologies. The residen- tial market for high speed packet-switched networks, for example, is obviously very small. Consequently, private line customers are essential testing grounds for new technologies which will later become part of carriers’ public network offerings. Although these two networked worlds often seemed sepa- rate--one public, the other private, one for voice, the other for data--they were (and are) engaged in a tightly coupled evolutionary dance. Public telephone operators (PTOs) own the principal undersea cable and satellite facilities which provide the backbone circuits for the world’s data networks, including the Internet. Private line customers provide a sub- stantial portion of the demand for this transmission capaci- ty, much of which is currently under-subscribed (see e.g., Figure 4). PTOs and private network operators thus have a shared interest in the terms on which international circuits may be leased and connected to the public switched tele- phone network (PSTN). PTOs depend on their private line customers in other respects. The multinational companies which lease most private lines also generate large volumes of public telephone traffic. If these customers are not well served, they may shift traffic to their own networks. PTOs also rely on private Still for much of the last decade, co-evolution has depend- ed on the fact that private and public networks primarily offered different services. Cooperation rather than compe- tition was the rule. But it is now evident that the environ- ment is changing again. The evolutionary strategies which allowed both networks to flourish in the past may no longer work. This time, though, in contrast to the early 198Os, reg- ulation and technology are bringing the landscape occupied by these two networks closer together. For instance, today more and more of the PTOs’ interna- tional private line customers also directly compete in pro- viding public switched telephone services. Some are mere- ly service resellers and are new to the market; others are PTOs which seek to use private lines as an inexpensive means of access to foreign markets. In addition, one of the largest new sources of demand for international private line circuits is from tntemet service providers (ISPs). And XII © TeleGeography, Inc. 1996 Introduction though the lnternet was built for e-rea!!, tomorrow it may carry telephony too. Technological innovations also have led PTOs themselves to compete directly with their private line customers. As the public network becomes more and more digital, telephone companies are seeking to provide a range of end-to-end data services, wooing small and medium sized customers who might otherwise lease their own lines. Many PTOs are also actively marketing Virtual Private Networks (VPNs) using "smart" switching and call routing technologies to par- tition the public networks on an as-needed basis for private customers. Some PTOs have become Internet service providers themselves and others are keen to join them; wit- ness WorldCom’s proposed acquisition of MFS/UUNet, now one of the largest ISPs in the U.S. and Europe. The data and articles presented in the body of this report are intended to flesh out the story of this co-evolutionary process. Will the current cycle of change lead to the end of public networks as we know them? Or will private networks begin to die out as PTOs bring their greater marketing power and facilities to bear?. Or will we see a third sce- nario-hybrid networks--as often occurs in nature when environmental changes force two major species to compete side-by-side in the same territory? Of course, these ecological metaphors are inexact. They are intended to suggest the dynamic process of change that is now underway within the international telecommunications business rather than to define it precisely. Other scenarios are also possible. Moreover, the process of change will be uneven, with significant variation from region to region and sometime even from country to country, although in the age of telephone call-back services and the Internet, no country can easily prevent its citizens from taking advantage of telecommunications services available elsewhere. These cautions should be kept in mind in reviewing the trends highlighted below. Industry Trends ¯ Volume--The volume of international traffic on the public telephone network continues to grow much faster than the global economy with traffic volumes up 15 percent from 1994 to 995 (see Figure 1). We expect traffic to grow at a similar rate (12 percent or more) in the next three to five years, generating over 106 billion minutes of international traffic by the year 2000. This expansion will be driven, in part, by the addition of over 425 million new telephone cus- tomers which, by 2000, will raise the base of telephone sub- scribers to 1.2 billion, including approximately :500 million mobile subscribers (see Figure 2). ¯ Competition--More and more national PTOs face global competition for their international telephone services, even where additional facilities-based carriers are not permitted. Telephone call-back and credit-card based services provided by carriers outside their home market generated at least three billion minutes of traffic in 1995. Most of this traffic was handled by U.S. carriers--even AT&T offers call-back services from Europe~which increased their share of the global market to over 25 percent in 1995 (see Figure 3). This was due in part to a large jump in resale traffic. U.S. resellers of international switched services increased their 1995 traffic base to 2.3 billion minutes from 1.3 billion in 1994. p: t or Be Eaten: North American Carriers are at the Top of the Global Food Chain .S. Carriers" Share of International Transit Traffic Carried by T~l~globe Telephone Traffic, 1986-1995 Canada, 1991-1995 1986 1989 1992 1995 1991 Transit traffic ~Percent transit 1992 1993 12% © Telel3eography, Inc. Xlll TeleGeography 1996/97 © TeleGeography, Inc. 1996 The impact of this U.S. assault is striking. For example, in 1995, on the U.S.-India route, the volume of telephone traf- fic billed by India’s overseas carrier, Videsh Sanchar Nigam, was essentially unchanged from 1994 at 52 million minutes whereas the volume of traffic billed by U.S. carriers increased over 50 percent to approximately 285 million minutes. The statistics tell a similar story on many other routes to and from the U.S. Traffic billed by U.S. carriers accounted for over 85 percent of the volume increase over 1994 on the Hong Kong, Japan and Argentina mutes. ¯ Toward Bill and Keep--New billing and routing arrange- merits for telephone traffic (e.g., call-back and home coun- try direct services) suggest that carriers which originate international calls are gaining the upper hand over carriers which terminate calls. To put it in another way, there is a new global call origina- tion business which is more and more independent from the call termination business. In the old world, these two busi- nesses were largely symmetrical and cooperative: the same carriers picked up and delivered calls. Routing and financial arrangements were mutually agreed; and carriers at each end of a route were paid the same amount for terminating a call. In the new world, call origination---control over the cus- tomer-is all important, and routing and settlement arrangements follow from there. This plainly calls into ques- Figure 4. How Do U.S. Carriers Use Their International Facilities? International Circuit Status of U.S. Carriers on Selected Routes, December 1995 Route Percent Idle Percent (to/from U.S.) IPLs PSTN IPL circuits idle Canada 5,543 44,172 11% 1,936 4% Mexico 1,653 23,416 7% 800 3% Hong Kong 860 742 54% 1,036 39% Japan 2,241 4,619 33% 16,259 70% Singapore 521 306 63% 593 42% United Kingdom 6,048 8,317 42% 27,001 65% All routes 26,497 126,150 17% 118,343 44% Note: C~rcu~t figures stated ~n 84 Kbps umts "Percent IPL" md~cates the share of total actzve c~rcu~ts that are privately leased and "Percent Idle" indicates the share of available czrcu~ts that remain unused PSTN refers to the public swrtched telephone network Canada Mexico Japan Singapore excess capacity XlV United Kingdom 0% 20% 40% 60% 80% Percent of Total 64 Kbps Clrcur~s ¯ IPL [] PSTN [] Idle mrcuits 1 100% J @ Tele6eography, Inc. 1996 © TeleGeography, Inc. 1996 Introduction F 5. The Top 20 Carriers: Traffic Growth Company Country 1993 1. AT&T United States 7129 2. Deutsche Telekom Germany 4680 3. MCt United States 2839 4. BT United Kingdom 2310 5. France T61~com France 2576 6. Telecom Italia Italy 1610 7. Swiss PTI Switzerland 1572 8, Sprint United States 1175 9. Hongkong Telecom Hong Kong 1377 10. Stentor Canada 1552 11. KPN Netherlands 1238 12, China MPT China 870 13. Mercury United Kingdom 820 14. Belgacom Belgium 979 t5. KDD Japan 952 16. Telef0nica Spain 847 17. Telmex Mexico 625 18. Austrian PTT Austria 767 19. T~l~globe Canada 808 20. Telstra Australia 640 Outgoing Traffic 1994 7947 5147 3517 2489 2603 1708 1649 1471 1578 1995 CAGR 93-95 8482 9.1% 5244 5.9% 4458 25.3% 2909 12.2% 2805 4.4% 1908 8.9% 1778 6,4% 1765 22.6% 1692 10.8% 1467 -2.8% 1459 8.6% 1339 24.1% 1107 16.2% 1106 6.3% 1086 6.8% 1025 10.0% 950 23.3% 901 8.4% 898 5.4% 807 12.3% 1525 1346 1090 1018 1049 1011 948 844 819 861 69O International traffic in MiTE CAGR is the compound annual growth. See page 64 for additional carriers and information. Inc.,1996 ©TeleGeography, J tion the survival of carriers which depend heavily on settle- ment revenue to make ends meet (i.e., from terminating more traffic than they send out). As yet, the redirection of customer traffic from one PTO to another is only marginally related to competition from International Simple Resale (ISR) carriers which lease dedi- cated international circuits in bulk and then break them down to provide retail telephone service (see generally pp. 15 to 21 below). However, our companion directory, New lnternutional Curriers, updated as of July 1996, shows that liberalized markets are now served by at least ]50 ISR providers. We expect major incumbent carriers (e.g., AT&T, BT, Telia, T616globe, Sprint) to begin using ISR facilities soon too, redirecting some traffic to affiliated carriers and "hub- bing" other traffic streams in a more cost-effective way to third countries. The distinction between originating and terminating carriers is also being driven by the Internet. Because the Internet is a connectionless service (data packets are sent from switch to switch without the need for a dedicated end-to-end cir- cuit), most Internet operators do not have close economic ties. Each commercial operator is responsible for funding its piece of the network regardless of the costs (or benefits) these investments bestow on the network as a whole. This is at once the beauty and bane of the tnternet today-- beauty, because central planning and traffic settlement regimes have been avoided--and bane, because without economic agreements between the Internet’s myriad access providers, there is little incentive to conserve the Internet commons for all. More on this below. ¯ Network Capacity--There is still substantial idle trans- mission capacity on major trans-Atlantic and trans-Pacific telecommunication routes. Circuit status data compiled by U.S. international carriers show that at December 1995 approximately 44 percent of the carriers’ available capacity was unused. On key routes, where new fiber optic cables are available (e.g., to the U.K. and Japan) unused capacity is even greater, averaging 65-70 percent (see Figure 4). The next generation of mega-cables now being planned for these routes will probably keep cable fill ratios at 50 percent or less well into the twenty-first century (see p. 63). These facts suggest that incumbent carriers (i.e., cable own- ers) will have more and more flexibility in the pricing and scope of their international offerings. Indeed, the magni- tude of the current gap between supply and demand for transmission facilities suggests that if international private lines (IPLs) were repriced a host of new international carri- TeleOeography 1996/97 © TeleGeography, Inc. 1996 ers and services--for Internet telephony, on-line music, cat- alog shopping--could be accommodated almost overnight. ¯ Private Lines--The foregoing circuit status reports also suggest that private networks have quietly become the largest users of international bandwidth on major routes. Simply put, data, not voice, is already driving the global network much more than many people realize. For exam- ple, 50 to 60 percent of the circuits between the U.S. and major Asian points (e.g., Singapore and Hong Kong) are pri- vately leased and over 40 percent of the circuits between the U.S. and U.K. are now dedicated to private customers. The preferences of these private users--whether it be trans- mission protocols or service suites, back-up capacity or pric- ing arrangements--are thus likely to have a large impact on the evolution of the public network. This can already be seen at the margins as major networking protocols, first used to interleave voice and data traffic on private net- works, are now being offered to the general public. The commercial initiatives of SITA, the network owned by the world’s airlines, form a case in point (see p. 12). ¯ lnternet Traffic--The mix of traffic on the Internet has swung decisively toward multimedia. Since early 1995, World Wide Web traffic has risen from 20 to 75 percent of total Internet packets (as measured by traffic over a major U.S. backbone network). The change is dramatic, and sim- ilar, in some ways, to the early years of radio, which rapidly changed from a ship-to-shore communications device to a mass broadcast entertainment medium. The new dominance of Web traffic on the Internet is proba- bly good news for PTOs. Though a multimedia Internet with real-time audio and video service presents a clear challenge to telephony, PTOs have the bandwidth and, in some cases, the networking skills and switching facilities to accommo- date it. Moreover, because the Internet is largely deregu- lated, the more popular it becomes, the more support PTOs are likely to gain for a more liberal approach to regulating plain old telephone service. This is especially true to the extent that the Internet provides an alternative dial tone. A viable Internet phone service must, however, surmount a number of hurdles (see pp. 57 to 40). ¯ Internet Metrics--Though the Internet works much of the time, performance is highly variable. Frequent users must now put up with dropped packets, unpredictable routing and even network brownouts. In fact, some experienced lnternet engineers, such as Robert Metcalfe, founder of 3Corn and the father of Ethernet, contend that the Internet’s unchecked growth may well lead to a catastroph- ic "crash." Though Metcalf’s forecast is flatly rejected by others (see p. 28), there is little question that the Internet’s service quality now falls short of that expected by telephone users. What is to be done? There are probably as many answers as there are lnternet engineers. To all we have but one response: whatever is done, the metrics may matter most. Without a cooperative method of traffic measurement, many technical solutions are likely to be ineffective or prob- lematic as users, ISPs and PTOs continue to argue over the results. Moreover, metrics are the key to developing pricing mechanisms that will help transform the Internet from an ad hoc private network of networks to a robust end-to-end communication service for the general public. Final Thoughts: the Language of Tomorrow No evolutionary survey would be complete without some attention to the language we use to describe the networked world. How we talk about communications networks-- whether we use organic metaphors (ecosystems) or geo- graphic ones (highways)--often reflects how we think about them (e.g., as in flux or relatively fixed). In turn, how we think about networks affects our expectations--what we want them to do today and in the future. Since the early 1960s, we have been constantly told that satellite communications and computer networks will make the world ever smaller--"a global village," in the words of of the late Canadian media scholar, Marshall McLuhan. Yet, as we first argued in TeleGeography 1993, instantaneous glob- al communication has not led to the end of geography so much as the explosion of place. This is largely a matter of economics. As the price of getting on the telecoms map has fallen, the demand for places has mushroomed. The global telephone network is now populated by hundreds of millions of people from Mexico to Malaysia and Finland to Fiji. In this bewildering new landscape, McLuhan’s metaphor has had a continuing appeal because of the way it evokes home and community. We have also used metaphors to tame the most unruly com- munications place of our time, the tnternet. Although cyber- space is anything but one-dimensional (see the End Page to this report), it is popularized as a hybrid form of real estate--an "electronic frontier," to be developed and mar- keted much like a new housing subdivision or shopping cen- ter Cyberspace may be a "land of bits" but it is a real, material one. And eminently bankable. Just ask the folks at Netscape, Yahoo!, Excite, Spyglass or the dozen or so other lnternet companies whose initial public offering of stock sold out in the last year. Despite our sympathies for this digital property boom (our Web site--http://www.telegeography.com--is for rent too!) there are competing, non-material metaphors for the Internet which may well be more illuminating in the longer term. For example, some writers, such as Finland’s Leena XVl © TeleGeography, Inc. 1996 Introduction untries Call Out, Poorer Countries Have a Traffic Surplus 6,000 International Traffic (millions of minutes) Incoming Outgoing 4,000 2,000 0 2,000 4,000 Sn Lanka Bangladesh Ecuador Uruguay Slovak Repubhc Pak=s~an Peru Cyprus Hungary Finland Braz~I Spain Japan Hong Kong Countries with a Traffic Surplus O O Countries with a Traffic Deficit 6,000 80% 60% 40% 20% 0% (20%) (40%) (60%) (80%) Surplus (Deficit) as a percentage of total traffic Note: Countries with a surplus receive more international traffic than they send. © TeleGeography, Inc., 1996 XVII TeleGeography 1996/97 © TeleGeography, Inc. 1996 Krohn, have depicted the Net as an "electronic oceanos°" of words and thoughts, like literature. The fact that precious few Web pages have literary merit does not make the metaphor inapt. Taken as a whole, the Web, like literature, makes the world bigger and smaller at the same time; it is also at once intensely intimate and incredibly public; part body, part mind, a parallel space for reflection and evolution. We live in a world of readers as well as builders and mer- chants. The inventors of Netscape NavigatorTM, the most popular software for "browsing" the World Wide Web, bor- rowed as heavily from the literary metaphor as they did from the real estate one. NavigatorTM starts at a "home page" and allows users to mark their way not by "road signs" but with "book marks." And why not? The hypertext transfer protocol (http) which underlies the Web was designed by Tim Berners-Lee in large measure to permit the collaborative review and creation of documents. So what? Why should we care about these metaphorical divides? Does it matter whether we talk about the Net as literature or as part of a ecosystem rather than as a market or a highway? Isn’t the Internet all these things and more? Perhaps. But, language does have consequences. If mate- rial views of the Net dominate our vocabulary, as seems true today, then many communication businesses are likely to flesh out the metaphor in their investment plans. The infra- structure installed, the look and feel of the access points, the types of services which are provided--all these are like- ly to reflect their investors’ preferred vocabulary. Geographical metaphors are likely to bring us more geogra- phy: wider and wider highways; ever faster modems and any number of electronic cars to drive. Yet, for most people, to borrow from Microsoft’s ads, the question will still be: "Where do you want to go today?" And "why"? On the other hand, if we think about the Internet (and other networks), in a more organic way, a far more cautious approach may be indicated. A highway is not a habitat nor is a book literature; each comes into its own in relation to the community or the society in which it was created. Few successful networks are autonomous; they are a piece of something greater than themselves and rarely blossom with- out the right social fertilization. A contrarian view of the information landscape---for exam- ple, as one shaped by browsers as much as by drivers--also suggests that the most successful communication business- es five years from now may be distinguished less by how many customers they have wired than by how many people they have taught to use the wires that others have installed. Companies that answer the "where" and the "why" for their clients may be treasured more than those which simply lay the asphalt or build the home sites. And when it comes to the "where" and "why," engineers may be less important than educators, and marketing skills less important than mentoring. Gregory Staple October 1996 Washington, D.C. XVII! Beyond the Public Switched Network TeteGeography 1996 © TeleGeography, Inc. 1996 © TeleGeography, Inc. 1996 Sizing the Market for International Private Lines SIZING THE MARKET FOR INTERNATIONAL PRIVATE LINES by Graham Finnie and Melanie Stockweli Though sometimes regarded as a distraction by telecom- munications operators (TOs), the market for internation- al private lines is by any measure large, growing, and profitable. In 1995, telecoms customers spent approximately $2.1 billion on international private lines (IPLs), representing roughly four percent of all international telecommunica- tions revenues (see Table 1). Of this, retail demand from private corporations for their own use accounted for approximately $1.5 billion. The rest came from wholesale buyers, such as lnternet Service Providers (ISPs), perhaps the fastest growing source of demand for dedicated international capacity, and from other resellers, global alliances and managed data network providers. (Details on the growth of glob- al Internet backbones can be found at pp. 53-55 below.) We estimate that the international private line market by capacity has grown at between 1 5 and 20 percent annu- ally for the past five years, roughly paralleling the growth in international switched minutes. However, prices for IPLs are falling faster than the switched network--at approximately 10 percent a year on major routes. Thus, IPL revenue is lagging circuit expansion, though there are regional variations. More importantly, despite the absolute growth of the IPL market, the proportion of corporate traffic traveling on private circuits is beginning to decline. We expect that relative decline to gather pace in the next few years as users make greater use of public services, including vir- tual private networks (VPNs) for voice, and frame relay and managed lnternet Protocol (IP) services for data. Initially, the decline in IPL usage have been confined to data communications, but it is now spreading to voice communications as well. However, this decline is relative: the explosion in data communications, brought about in large part by the shift to client-server computing, is benefitting all telecom set- vices, including private lines. Traditionally, computing architec- tures were based on dumb termi- nals linked to smart mainframe computers that were easily (and most often strictly) controlled by centrally located information technology (IT) staff. By contrast, client-server computing, using personal computers (PCs) with off-the-shelf software, is largely driven at the departmental and divisional level. The new networks have created a near-anarchic situation: corporate IT staff in some big companies now have only the barest idea of who is transmitting what data and where. Table 1. 1995 International Telecommunications Revenue Service Revenue PSTN $55 billion Private Lines $2.1 billion 96.3% 3.7% Note: Revenues for Managed Data, VPN and resale have been exclud- ed since th=s would result m "double counting" of revenues. Private hnes are used for all three services, and the PSTN is also used by some resellers As we note in our two case studies, (see pp. 1 2 and 13 below), client-server computing has helped create unprecedented demand for all forms of wide area com- munications, and now the Internet and its associated technologies are adding to the swelling rivers of data. Applications that were formerly confined within buildings and departments have spilled out across national bound- aries; and those applications themselves are continually evolving in ways that quickly turn bit streams into tor- rents. For example, e-mail (once a very modest consumer of bits) has been replaced by client-server-derived e-mail with attachments (e.g., Lotus Notes), an increasingly large consumer of bits, and an increasing problem for telecoms managers. The same can be said for lnternet software which initially allowed little more than plain text to be communicated and now allows users to send almost anything that can be digitized. The fact that indi- vidual workers can themselves download application software off the Internet adds to IT woes. Graham Finnie is Research Director and Melanie Stockwell is a Research Associate at the Yankee Group Europe (YGE). This article is adapted from a forthcoming YGE report on global private line networks. Contact YGE at: (tel) +44 I923 24 6511 (fax) +44 1923 24 2456 For providers, the result is a continuing data communi- cations bonanza. The client-server revolution was driven from the bottom up, and its impli- cations for wide area communica- tions are still being worked through; the related boom in lnternet communications has only just started. TeleGeography 1996/97 © TeleGeography, Inc. 1996 Much Do Leased Circuits Cost? s International Circuit Prices AT&T KDD US-Japan AT&T Telecom Italia US-Italy AT&T TelefOmca Telia Swiss Telecom Sweden-Sw,tz BT France Telecom UK-France ~~ 40,000 80,000 120,000 Total Circuit Coat IUS$) per Month 160,000 64 Kbps International Circuit Prices AT&T HK Telecom US-HK ~’ o AT&T KDD AT&T Telecorn Italia AT&T Telof6ni~a Telia Swiss Telecom Sweden-Swrtz ~’~.~ BT France Telecom UK-Fran~a ~,.’,~ 0 4,000 8,000 12,000 Total Czrcuzt Coat IUS$) per Month 16,000 Generally speaking, prices have fallen by about 30 percent since mid-1993, continuing a trend that has persisted for at least s~x years. However, this generalization conceals a great deal of local variation; for example, Deutsche Telekom and Telenor (Norway) have cut tariffs by up to two-thirds since 1993, but France Telecom and Telecom Finland have main- tained pricing at 1993 levels. In a few ~nstances, tariff structures have changed: Deutsche Telekom, for instance, now charges separately for national extensions on a case by case basis (unlike most other carri- ers), so the precipitous fall in pricing shown in the table is a httle m~sleading. These differences also often reflect underlying commercial or political strategies. Telecom Finland, for instance, has placed heavy emphas~s on managed data services, and may have retained relatively higher prices to encourage migration of users to this service. There may also be very wide variations in pricing. While OPT of Austria and Telefonica make little differentiation between the cost of a circuit to the nearest European country and to Japan, the differential for Telenor is 15:1. The following table provides indicative figures on the decline, in real terms, of IPL prices from 1993 to 1996. Monthly rental for a 64Kbps half-mrcuit in 1996 as a percentage of the 1993 price (1993 price = 100). All prices are m US$ to near to far to to Country Europe Europe USA Japan Austria 64.1 63.2 58.8 44.1 Belgium 53.1 84.0 51.3 70.1 Denmark 94.9 94.9 90.7 NA Finland 86.8 94.3 89.6 96.0 France 94.3 94.3 81.6 NA Germany 37.3 37.3 39.2 47.4 Greece 56.1 58.9 45.5 45.5 Ireland 39.2 65,2 74.5 93.1 Italy 75.0 68.5 79.6 71.8 Netherlands 52.7 59.5 59.7 63.0 Norway 23.9 36.0 41.7 68.5 Portugal 84.4 80.2 71.5 76.7 Spain 64.6 64.6 74.2 74.2 Sweden 68.3 76.7 81.8 NA Switzerland 80.5 76.2 80.0 72.0 U.K. (BT) 84.2 44.7 87.9 88.1 Average 66.2 68.6 69.2 70.0 © TeleGeography, Inc. 1996 What are International Private Lines? Private lines are ordinary telephone circuits that are leased for a flat fee to a particular customer and thus always "open." A private line, in industry jargon, is "nailed up" by the TO and bypasses standard switching equipment. Originally, circuit cross-connections were hand-wired; since digitization, this is usually accom- plished electronically using digital cross-connect equip- ment. An international private line requires the lease of two matching national circuits for the provision of through service. Typically, the customer must deal with two dif- ferent national carriers although, as discussed below, a number of joint marketing arrangements and carrier alliances now permit customers to provision whole cir- cuits through a single carrier contact point. Customers pay a connection charge (though this is increasingly waived in competitive markets for all but the shortest © TeleGeography, Inc. 1996 Sizing the Market for International Private Lines contracts) and a monthly fee for use of private lines, but no volume charges. The use of private lines by private corporations--as well as presidents and prime ministers--has a long history. The first private lines were installed soon after the tele- graph was invented, and before public monopolies were created, by companies such as Reuters (a leading news wire service). Over the past two to three decades, the use of IPLs has spread rapidly as companies have sought reliable connections between remote computers. For most of the 1980s, many TOs, especially in conti- nental Europe and Asia, tried to persuade corporations to stop using private lines and use instead the TOs’ pub- lic service alternatives, which included X.25 packet switching. TOs argued that IPLs were a crude historical legacy which made wasteful use of scarce capacity (since they often go unoccupied for long periods). In a digital age, they reasoned, where network capacity can be allo- cated flexibly, public service is both more reliable and more efficient. In some countries such as Germany, the TO began to charge for volume and usage on leased circuits; Germany and one or two others also prohibited outright the trans- mission of voice on leased circuits, while most European and Asian TOs prohibited any breakout of voice--and Box 2. Five Percent of the World’s Telephone Traffic is Carried by Private Networks How many minutes of international voice traffic travel over private networks, as compared to the public network? We know the latter figure to a high degree of accuracy: totalling the statistics in this report we reach a total of 60.3 billion minutes for 1995. Resellers using IPLs ft. e., international simple resale (ISR) carriers), rather than carriers reselling switched minutes, probably account for another 800 million minutes. The bulk of ISR traffic travels to and from the U.K. or on the U.S./Canada route (see p. 19 below). However, calculating min- utes of voice traffic on private lines used by private organizations is much harder. Most suppliers don’t know how their customers use private lines--and, worse, many network managers don’t monitor line usage in detail. Nor are industry-wide statistics available on how "full" those lines are; though we know that private lines usually are not cost-effective for voice alone unless they are used three to five hours a day. But for IT managers, voice often "rides free" on networks installed to handle vital data traffic. Equipment can route voice over the private network when there is surplus capacity, but this is rarely mea- sured in detail. Moreover, where voice is compressed to 8Kbps or less (as it routinely is on many private cir- cuits), the crossover point for using a leased circuit may be much lower than three to five hours daily. Lesser but significant problems include the absence of precise information on exactly how many private lines there are (see main article), and how many of these are used by private corporations. Our estimate thus makes the following assumptions: ¯ 25,000 international private lines (whole circuits) in use by private corporations (out of a globa! total of 36,000--see Table 4), with an average capacity of 64Kbps (this takes account of the fact that there are a significant number of analogue lines with lower capacity and a small but important proportion with higher capacity; our margin of error on this f/gure is l 0 percent either way); ¯ Five voice channels per 64 Kbps circuit, reflecting standard compression to 8 or 16 Kbps (the true figure may be anywhere between four and six channels in our view); ¯ Four hours average utilization per circuit, of which an average 1.5 hours is voice (the actual figure may be as low as one hour and as high as two hours); ¯ circuits are utilized 240 days per year This yields the following calculation: 25,000 x 5 x 90 x 240 = 2.7 billion minutes, or 4.5 percent of aft telephone traffic (give or take one percent). This calculation does not exclude fax. As we note in the main article, the proportion of fax traffic in the inter- national PSTN is thought to range between 30 and 60 percent, depending on the route. The share of fax traffic is much lower in private networks--no more than ten percent. Conventionally, fax traffic is routed even by large corporations on dedicated PSTN circuits because it is difficult for technical reasons to route it via the PBXs and multiplexers that feed into leased c~rcuits. © TeleGeography,JInc.1998 5 TeleGeography 1996/97 © TeleGeography, Inc. 1996 sometimes data--into the public network. But corpora- tions stuck doggedly to their view that, in many situa- tions, a pdvate line the made most business sense. The Demand for IPLs Corporations lease international lines for three main rea- sons: ¯ First, if the volume of traffic between two points-- two major corporate sites, for example--is sufficiently high, then it is cheaper to use a private line because there are no volume charges on private lines. Historically, this crossover point has been three to five hours a day, though it has varied. TOs have often raised prices to discourage IPL use. Conversely, others have lowered rates to encourage companies to route private lines through their TO’s national hub, which may provide an opportunity for the TO to transit and refile (to off-net locations) a portion of the corporation’s traffic. ¯ Second, corporations use private lines because they have historically been more reliable than switched con- nections (the service quality can be directly monitored and maintained), and for certain applications this has been paramount. Two examples are: the links between financial trading rooms, where failure can be disastrous- ly expensive; and links between computers, which on many routes until this decade could only be realized reli- ably over private lines. For critical applications, users built meshed networks in which there were two alternate routes to key sites. ¯ Third, some applications simply cannot be operat- ed over switched facilities. A good example is remote front-end processor (FEPI operation, whereby a front- end processor--which is normally situated within a few meters of a mainframe--is situated in another country Table 2. Top Ten IPL Routes Number of lines, all types of circuits; all estimates Route Lines U.S.-Canada 7,000 U.S.-Mexico 1,500 U.S.-U.K. 1,500 U.K.-France 1,000 Germany-France 800 Germany-Switzerland 700 U.K.-Germany 700 U.S.-Japan 700 U.K.-Netherlands 650 Netherlands-Germany 500 Note: These esttrnates are for mrcu~ts of varytn~I capacity {see also Box 5 onl) 19 for 1995 data on 64 Kbps IPL c~rcu=ts on selected U.S routes). Source: the Yankee Group Europe, 1996 and connected via a "channel-connect" facility. The bandwidth required exceeds anything that can be pro- vided continuously by any dial-up services, and often the delays inserted by managed services are too great to ensure proper operation. For users, the benefits of oper- ating a remote FEP include the cost reduction associated with the removal of mainframe computers in remote countries, and the economy of centralized IT control. Competition among equipment vendors such as Micom, NET, Newbridge Networks, Stratacom, Tellabs and Timeplex (now Ascom Timeplex) has generated a wider and wider variety of high quality equipment for econom- ic utilization of private lines. Multiplexing equipment, for example, makes it possible for a user to obtain up to eight voice channels from a given 64 Kbps circuit. Corporations may now compress voice to 8 Kbps or less, lowering the crossover point at which it becomes eco- nomically attractive to shift traffic from the public net- work. At the same time, however, the service offered by IPL suppliers has greatly improved. Many now offer guaran- tees on their half of the circuit, with stiff penalties if they are not met. BT, for instance, pays a rebate of 30 per- cent of the line cost if it is out of service for six hours or more per month. And the time for provisioning circuits, a big problem in the past, is gradually improving. IPL prices are falling too. The price of a 64 Kbps half-circuit has fallen about 30 percent in the past three years, and fell 30 percent in the previous three years as well; greater savings are possible if new discount schemes are taken into account (see Box 1). As a result, private lines now are very widely used by major corporations: in a survey of 50 large users under- taken in mid-1996 for this article, the Yankee Group found that 54 percent were using IPLs; and among large international corporations, all were using tPLs. In a larg- er survey of 150 users in 1995, a similar proportion (51 percent) said they used private circuits. We estimate that there are approximately 56,000 IPLs (whole circuitsl in use worldwide by private corporations, totaling perhaps 2.2 Gbps of total capacity. As a pro- portion of all telecommunications capacity, this is rela- tively insignificant, representing approximately 5 percent of voice traffic (see Box 2). But for the corporations con- cerned, this capacity has been a vital asset in their effort to create more cost effective and more reliable commu- nications among far-flung sites. Indeed, according to suppliers and our own user survey i data, many corporations are adding further sites to exist- ing networks, or increasing the bandwidth used on exist- © TeleGeography, Inc. 1996 Sizing the Market for International Private Lines I .Table 3. Corp.orate Demand for International Services (1996) Based on YGE telecom manager opinion poll 1. How is your international voice traffic carried? Reseller 4% Private Network 9% IVPN 14% PSTN 73% 2. How is your international data traffic carried? Public Network 16% Managed Data Network Service 36% Private Network 48% Note: Figures represent average percentage of aggregate traffic, not j~ercentage of users using each service. Source. Yankee Group ~-urope, ing routes. As multinationals expand, traffic to remote sites increases, justifying the installation of private lines on cost grounds; elsewhere, the relentless increase in data traffic (typically 20 percent or more a year in the average corporate network) drives the need for more capacity on major routes. Most large international corporate networks were based on 64 Kbps lines until very recently; now, the use of frac- tional T-1 and E-1 lines--unavailable on most routes until recently--has become widespread. A few major corpora- tions are using full T-1 (1.544 Mbps) or E-I (2 Mbps) lines. The Yankee Group estimates that there are about 200 to 300 T-I or higher capacity IPLs in use today by private corporations. Where traffic justifies it, these lines can be quite cost effective: E-1 lines, for instance, which carry 30 64 Kbps circuits, are priced at nine to 15 times the cost of a 64 Kbps circuit, depending on the supplier. Meanwhile, a very small number of corporations have made the move to even higher-speed lines, and others are considering it. This year, for instance, Hewlett Packard, a major corporate network operator (see Box 4), issued a request for tender for multiple T-1 capacity on the trans-Pacific route. However, few corporations can yet justify such high speed lines, and most of the demand at these upper levels is coming from Internet and other on-line service providers. How IPLs Are Used Table 3 shows the proportion of international corporate traffic carried by different TO services based on our sur- vey of approximately 50 major companies. TO estimates and our own data on the split between data and voice on IPLs is somewhat contradictory. It is clear that data predominates, but it is not clear by how much. Our own view is that about 65 percent of traffic by cir- cuit occupancy is data, about 35 percent voice. The most important routes for private circuits roughly track those for switched minutes (see Table 2). For example, routes between the U.S. and Canada, Mexico and the U.K. occupy the top three positions just as they do for switched minutes. However, there are some important quirks. As Table 4 shows, for example, Germany and France lag well behind the U.K. in total circuits installed, though Germany gen- erates significantly more public international traffic than the U.K. The reasons for this are primarily historical; the U.K. has always been an important hub for international private line traffic; it is the most important financial cen- ter in Europe; and the use of private lines has never been discouraged there, as it was in many continental European countries. Prices have generally been much lower too. In other cases, a concentration of large corporations or financial services companies makes for a large market, as in Switzerland. Countries like Belgium are important hubs and transit points. Another obvious difference is that private line traffic is business traffic, whereas public Table 4. Estimated IPLs, by Country IPLs for all users as of January 1996 Country Lines Australia 1000 Belgium 2200 Canada 8000 Denmark 500 France 3500 Germany 6000 Hong Kong 1500 Ireland 600 Italy 1600 Japan 1800 Netherlands 1400 Singapore 1000 Spain 1000 Sweden 750 Switzerland 1700 U.K. 11000 U.S. 17000 Rest of World 10450 Total 71000 Note: Data are for all mrcu=ts leased to pnvate corporations and resellers ~nclud=ng analogue. Higher speed c~rcu~ts (eg 512Kbps 2Mbps are counted as one circuit, Note that total number of whole c[rcu=ts =s approximately 36,000 since two half c=rcuits are requ=red to prowde through serwce see also Rgure 4 at p x=v for 1995 data on 64 Kbps IPL c=rcu~ts on selected U,S. routes). Source: Yankee Group Europe © TeleGeography, Inc. 1996 I TeleGeography 1996/97 © TeleGeography, Inc. 1996 pTable 5. Major IPL Suppliers Estimated 1995 Earnings Carrier IPL Earnings (US$) Cable & Wireless 200m AT&T 200m MCI 150m Deutsche Telekom 130m Stentor 125rn WorldCom 120m KDD 110m BT 110m France Telecom 65rn Sprint 45rn Source: Yankee Group Europe, 1996 network traffic measures business and residential traffic; where access to IPLs is limited, more business traffic will flow over the public network. IPLMarket Trends Despite the evidence that the demand for IPLs is still ris- ing, the market is a mature one and, relatively speaking, it’s on the slide. The long effort by suppliers to woo users onto public or "virtual public" alternatives is at last beginning to pay off. Carriers are lukewarm about pri- vate lines for a variety of commercial reasons, as well as the engineering prejudices mentioned above. First, the flat rate tariffing used for private lines runs counter to TO prejudices in favor of usage-based tariff- ing. When traffic is rising and the meter is running, TO’s reason, tariffing by use equals big margins. A related rea- son is that IPL traffic bypasses the accounting rate sys- tem. Thus, instead of receiving a per minute charge for landing traffic, the TO only receives a flat monthly fee. That is why, of course, TOs have tried to limit customers from reselling the capacity of IPLs to third parties for switched services (known as International Simple Resale or ISR) so as to stem a wholesale diversion of settlement ~venues. Second, most suppliers believe that the margins on ordi- nary IPLs could soon decline to a point where it will be difficult to make much money. The reason: it’s hard to differentiate or add value to such a simple product. "Private lines are as close to a commodity as you can get in telecommunications," a senior manager with Cable & Wireless told us. On most routes, margins are actually very good, because providers still monopolize or effectively control the instal- lation of cross-border capacity. But on the biggest and most competitive route, across the Atlantic, where there is considerable excess undersea cable capacity, margins are already under pressure. Although no TO will reveal financial figures, several told us that if pdces dropped much further, they would be making a loss on IPLs, espe- cially at higher speeds. That claim must, perhaps, be taken with a pinch of salt. According to the U.K. regulatory agency Oftel, the esti- mated annual cost to BT of providing 2Mbps circuits for wholesale use is £ 160,000, but this takes into account all BT’s historic costs; for a new operator self-providing circuits by buying ownership rights in a cable (so-called IRUs), Oftel estimates the cost would be £94,000 (see Box 3). In the table, Oftel calculates the cost in two ways: ( 1 ) the cost to a new operator of self-providing a circuit by buy- ing an 1RU, plus maintenance costs and local access cir- cuit costs; and (2) the cost to BT based on fully allocat- ed costs, overheads and so forth. By comparison, BT’s retail price for the circuit without any discounting is £244,000 (see also Box 3). This sug- gests there is still quite a bit of room for prices to fall, especially as new players and infrastructure providers begin to size up the market. All three major alliances (Concert, Global One and Unisource/Uniworld) as well as Cable & Wireless have launched "managed bandwidth" offerings, in which they take responsibility for end-to-end supply of leased lines. In principle at least, this removes the administrative has- sle from the user of having to buy from two or more sup- pliers. Most managed bandwidth providers also offer some level of guarantee on provisioning time and circuit up-time, with more or less severe financial penalties if they fail to deliver TOs are not keen on these offerings and often don’t advertise them. Apparently it’s hard to make much money supplying managed bandwidth. Suppliers must lease the access line wherever they don’t directly supply such lines, usually at retail prices, and this often wipes out profit on the line. Some local operators deliberately structure pricing to discourage managed bandwidth providers, whom they see as a threat. Carriers under pressure from competitors have in many cases widened the differential between 64 Kbps and 2 Mbps circuits so that it’s more difficult for competitors to buy wide band circuits, split them up for customers and still make a profit. In 1989, before competition began in Europe, the ratio between 64 Kbps and 2 Mbps circuit tariffs was about 1 : 10; today it is about 1 : 12, even though most cost data © TeleGeographyo Inc. 1996 Sizing the Market for International Private Lines suggests that the ratio should be falling. Making matters worse, the belief among suppliers that they could charge more for managed bandwidth than for leased circuits has been largely proved wrong. Most users will not pay more for managed bandwidth, and according to Global One, many actually want to pay less. Among customers, meanwhile, there is a growing desire to move to public offerings, driven in large part by senior management pressure to outsource everything that is not considered core to the business. This means that there has been strong pressure on telecommunications man- agers to justify the existence of a private network, which requires staff and management; in the Yankee Group’s 1995 user survey, we found that only one-quarter of users said they would definitely not outsource their cor- porate network. Whereas leased lines are effectively neutral as to traffic type, managed or public services are usually specifically designed for particular types of traffic. The next two sec- tions, therefore, look separately at developing demand for voice and data public network alternatives to IPLs. The Rise of Managed Data Services In data communications, there has already been a strong move to managed data offerings, and we expect that to continue. Our 1996 survey found that users expected to move away from private lines and towards managed data offerings, confirming a trend shown in a similar survey taken in !995. Suppliers say that the trend is a long-term one. Cable & Wireless, for instance, estimates that, while its revenue from IPLs has grown by about six percent per annum over the last six years, revenue from managed data offer- ings (including X.25, IP, and frame relay) has grown by 30-40 percent per annum. Other suppliers report simi- lar changes. There are several reasons why the move to managed offerings is happening first in data communications: ¯ The market was deregulated earlier. In most of Europe and Asia, deregulation of data communications service took place at least five years ago, meaning that the supply industry (and its offerings) is now relatively mature. For example, there are at least a half-dozen operators offering international frame relay services that reach ten or more countries (AT&T/Unisource, CompuServe, Concert, Global One, IDB Worldcom, Infonet). ¯ The pattern of corporate traffic and demand is changing. In the past two years, the rapid spread of client-server networks and LANs has spilled over into the wide area where it is now the main source of new demand (see also Box 5 regarding the rise of traffic in SITArs glob- al network). This type of traffic differs in two ways from traditional data traffic. First, it is more "bursty" (i.e., there are more peaks and troughs in bandwidth demand) which means that it is not very well-suited to leased lines where capacity is permanently assigned. Second, wide area networks (WANs) are less hierarchical Box 3. Offers Cost Equation Data on the true cost to operators of providing telecommunications serv=ces is notoriously controversial, since cost data is often not avail- able and few economists agree on how costs should be allocated. Nevertheless, virtually everyone agrees that the cost of supplying capacity is still much lower than the price charged. One of the few independent agencies that has attempted to assess the true costs is Oriel, the U.K. regulatory agency. As part of a consultative document issued in July 1996, Oftet came up with the estimates shown here. Sdf-provislon ioo ..... o~.. ,-,4,~;;;’s,..: 94,000 0 50,000 100,000 150,000 C~rcult Cost (£) 200,000 250,000 Oftel has proposed that the second cost calculation or "BT cost plus" wholesale rate he available to relevant connectable systems. These systems include resetlers and managed network providers but not ordinary private line carriers. Note: BT "cost plus", includes fully allocated costs, relevant overheads and an (unspecified) rate of return on capital employed Self-provision based on IRU cost maintenance retail 2 Mbps backhaul circuit charges (i.e., national extensions) Source adapted from Oftel data. @ TeleGeography, Inc. 1996 TeleGeography 1996/97 © TeleGeography, Inc, 1996 Box 4. The World’s Largest Private Networks Who has the biggest IPL networks and how big are they? The problems start with definitions--and there are as many as you like. What counts as a private network? SITA is a consortium that sells its services back to its airline owners. SWIFT has a similar identity in the financial services industry. In the end, we included in the listing here only those companies that built networks for their own use. SITA, however, which runs the largest non-TO network, is profiled in Box 5. And just how do you define "biggest"? One way is by measuring traffic, but few users have anything but the most rudimentary idea of how much traffic they generate, or whether it’s voice or data. And different net- works (e.g. circuit and packet switched networks) measure traffic in different ways (in this case, by minutes and data segments, a variable and rather esoteric measure). Indeed, poor auditing of network traffic is one reason why corporations often find it difficult to issue outsourcing tenders. Often there may be good infor- mation locally, but it is not consolidated globally. Budgets are a more objective measure which usually yield more reliable data--but there are almost insuper- able obstacles here too. The biggest is deciding what’s included and what’s excluded. Do you include some or all domestic communications or segments (some corporations don’t segment that out)? And all PSTN com- munications? What about equipment and depreciation? Even if you can establish an "objective" measure, most companies have their own way of assessing spending, and won’t have the figures you want. Moreover, many regard telecoms spending as a commercial secret. Taking into account the above reservations, the Yankee Group drew up a tentative list of the 30 largest ]PL networks using the following information: ¯ YGE aggregate data on corporate global telecom spending by industry sector and size; ¯ published information on spending by sector and size; ¯ published information on IT spending by individual companies; ¯ published information on major multinational companies (i.e., identifying how "multi- national" or transnational they were); Principal secondary sources were Communications Week, Communications Week International, Datamation, Information Week and the United Nations Conference on Trade and Development (UNCTAD). Some additional information was drawn from the U.K. regulatory agency, Oftel, and from European Commission databases. Thus this list is not the 30 biggest private network operators; but we do think it includes the top 12 spenders. How extensive are their networks? Some are very extensive. In the Yankee Group’s 995 User Survey, we asked 143 users how many countries were connected to their private wide area network. Of the 75 who actu- ally had an international WAN, 19 connected 20 or more countries, and six connected 50 or more. The largest connected 100 countries. Such networks don’t come cheap. With 64 Kbps IPLs averaging approximately $30,000 per annum per half- circuit, a large private network with aggregate capacity of, say, 10Mbps will cost around $10 million; man- agement, manpower, multiplexers and other gear will at least double that figure. The largest network owners, including all those on this list, spend at least $50 million a year on telecommu- nications, of which at least one-quarter is spent on IPLs. It’s worth noting that all these companies are dwarfed by the largest customer for international bandwidth, SITA. They will also likely be dwarfed in the near future, at least in terms of bandwidth, by Internet service providers, who are the fastest-growing consumers of international private lines. J © TeleGeography, Inc. 1996 © TeleGeography, Inc. 1996 Sizing the Market for International Pdvate Lines I Box 4. (continued) Who is most likely to have a big private network? Our list can be divided into five major categories: Major financial services companies--As we have noted elsewhere, these companies drove the early demand for private circuits, and they remain the most important category of IPL customer. Anecdotal information suggests that Citicorp and American Express, for instance, have among the largest international private networks. However, their importance has declined relative to other sectors: IT companies--It’s perhaps not surprising that IT companies should feel comfortable with large computer networks. Companies like Hewlett Packard, IBM and Siemens run among the very largest corporate networks; all three are almost certainly among the top dozen private network operators worldwide. Multinational conglomerates--A good example here is ABB, a company that drew together the disparate elements of Sweden’s ASEA and Switzerland’s Brown Boveri. ABB spends close to $300 million on communications per annum. Oil companies--Oi! companies have large networks partly by virtue of their sheer size, but also because they are highly global- ized. As a proportion of revenues, their spending on telecommuni- cations is generally no higher than average, but their dominance of the upper echelons of the Fortune 500 ensures their appearance here. Automotive companies--Again, size and global reach are major factors here. Some auto companies such as Ford have consciously globalized functions such as design engineering, greatly increasing network traffic. Some companies that might have been expected to creep into our list, such as the giant consumer goods concerns Nestle and Unilever, are absent because they do not operate private networks (both companies outsourced several years ago). Also missing are the world’s major airlines, which of course depend heavily on SITA for their network needs. 30 Major International Corporate Networks ABB Switz./Swed. Alcatel France American Express U.S. BP U.K. Citicorp U.S. Daimler Benz Germany Deutsche Bank Germany Digital Equipment U.S. Dow U.S. Du Pont U.S. Elf France Exxon U.S. Fiat Italy Ford U.S. General Electric U.S. General Motors U.S. Glaxo U.K. Hewlett Packard U.S. HSBC U.K./H.K. IBM U.S. Merrill Lynch U.S. Motorola U.S. Philips Netherlands Schlumberger U.S. Shell U.K./Neth. Siemens Germany Texas Instruments U.S. TRW U.S. U BS Switzerland Volkswagen Germany Note: Thzs listing is intended to be indzcatlve, not defimtive We believe that the top 12 private networks are all on th=s hst, but do not clmm that it ~ncludes all the top 30 private networks. Source: the Yankee Group Europe. © TeleGeography, Inc. 1998 than traditional dumb terminal networks that drove ear- lier demand. In such a network, data from one or sever- al terminals is collected and transmitted to some central site for processing, and is both predictable and control- lable. Whereas traditional networks have a star topology linking a head office or data centre with subsidiaries which is well-suited to leased lines, newer networks have many more point-to-point routes which are less heavily loaded--lending themselves to public solutions. In the last two years, demand for frame relay offerings has been especially buoyant. Frame relay is a variant on packet switching which is beginning to replace the public X.25 services widely used for international data commu- nications, especially in Europe. Frame relay gives users a so-called "Committed Information Rate" (e.g., 64 Kbps) for little more than the price of a leased circuit solution, but at off-peak periods, users may get a lot more than 64Kbps. IPL veterans say that actual through- put is not as high as advertised, and say that it’s not as TeleGeography 1996/97 © TeleGeography, Inc. 1996 Box 5 -The Big Network: SITA’s 1200-Node Monster The SITA network is a network of superlatives. With a total of 1200 nodes in 220 countries, it is the world’s most extensive private net- work, and with an aggregate network capacity of over 350 megabits per second, it is also probably the largest. More surprisingly, it may also be among the fastest-growing: in 1995, network traffic quadrupled, and in 1996 it was expected to triple-- an extraordinary rate of growth for such a large and relatively mature network. Owned by the world’s airlines and headquartered in Paris, the Soci6t~ Internationale de T616communications A6ronautiques (SEA) was constituted as a cooperative in 1949 to meet the airlines’ growing need for international communications. It has 600 airline members, and had a turnover in 1995 of over $1 billion. Huge Growth Today, network managers are grappling with a growth in demand that took nearly everyone by surprise. In the early 1990s, network traf- fic growth per annum was a relat}vely steady 30-50 percent; then in 1994, demand doubled, and it reached 400 percent in 1995. It’s a testament to the skill and experience of SITA’s management that it did not fold under the pressure; careful use of compression and bet- ter network optimization (achieved in part through the use of more "intelligent" routing equipment) meant that capacity requirements only doubled, for instance. The key reason for traffic growth is the same for SITA as for everyone else: the spread of client-server LANs and the proliferation of band- width-hungry applications running over them and out into the wide area. Originally, the SITA network was unique, carrying a mix of pro- pdetary airline protocols and applications that had their own particular characteristics. Now, the network looks much like everyone else’s-- except, of course, in its reach. Little voice is carried, but following deregulation on major routes, even that is now changing. Meanwhile, the network has been steadily snaking up river and inland; where before it was largely a one-node-to-one-country affair, the number of nodes has tripled over the past five years, and in the past 12 months, SITA has taken over large domestic networks previous- ly operated by British Airways, Lufthansa (Germany) and American Airlines, among others. For those running the network, therefore, life has been dominated by the need to get new capacity into service as quickly and reliably as possible. Between the 15 major nodes in the network, capacity is now at least 1.544 megab~ts per second (known as T1); most other links are now digital 64Kbps or multiple 64Kbps lines, though certain less-developed regions--most of Africa, for ~nstance--are st~ll large- ly analog. With capacity demand eating up more and more of the operational budget, a major objective of SITA has been to achieve better utiliza- tion of existing lines. Its working objective is that at any time 70 percent of circuits will be "normally loaded"; this is defined as 40-60 percent full on average, meaning they will be fully loaded at peak periods. In mid-1996, 60 percent of lines were normally loaded. Given that network traffic has become ~nherently more unpredictable and "bursty" due to the supersession of airline protocols by genedc LAN applications and protocols, that is an especially noteworthy achievement. Growing Cost SITA does not publish figures on the cost of its network, but a conservative estimate would be that it costs up to $1 O0 million for lines alone; with 5,000 staff worldwide, ongoing operational costs will be higher, and it is also spending heavily on new capital equipment The workhorses of the network are DPN 1 O0 packet switches from Nortel, which also provides DMS 1 O0 voice switches and Magellan switch- es. Cisco provides reuters. In an ever-changing network environment, SITA often appears a beacon of stability; but under pressure from its owners it’s being forced to change fast. Because of the replacement of airline protocols by generic communications and computer protocols, it is now implicitly competing against every other provider of data communications services--one reason why it decided to fight back and start providing services to non-airline users itself. In 1990, it established a subsidiary, Scitor, which sells services to all comers, including, most recent- ly, VPNs for voice. Though Scitor remains small, it is growing very fast Recently SITA received a S300 million cash injection from Morgan Stanley, helping to boost its commercial expansion. .J © TeleGeography, Inc. 1996 Sizing the Market for International Private Lines Fa Westinghouse: A Middle Sized Corporate Network go, the giant American energy and manufacturing conglomerate Westinghouse had a single 9.6 Kbps private line linking its U.S. and European offices. In the European morning, when U.S. staff were still in bed, it carried data between IBM mainframes, along with a little e-mail. In the afternoon, it switched to voice. Today, the company has ten sites connected at speeds ranging from 64 Kbps to 256 Kbps, making it a fairly typical medium-sized inter- national private network. Traffic is balanced approximately 50-50 between voice and data, with data applications dominated by e-mail, groupware, Internet access, and various engineering applications, often with a high graphical content. Each big project le.g., a new power station) creates a large graphical library that must be replicated at several sites, requiring that large files be routinely sent to synchromze these libraries. The Westinghouse network has followed a classic trajectory; higher speed analog links were followed in 1989 by the first digital circmts, which were exploited using time division multiplexers that offered a lower cost per bit. Like many other major corporations, especially those based in the U.S., Westinghouse had simply extended a model that had been wide- ly established in its domestic network, in which newly available digital leased lines were rented to improve the reliability of data transfer and, in some cases, cut the cost of voice communications between major corporate sites. With traffic largely flowing between the two headquarters sites m Pittsburgh and Brussels and out to regional offices--a classic "double star" network--the use of leased c~rcuits was especially attractive. Furthermore, leased circuits became increasingly reliable, which was a key concern for vital corporate data traffic. Westinghouse looks for 99.7 percent rehability on its internatzonal links, equivalent to 15 minutes loss of service or less per month; it rarely achieved that in 1990, but today it routznely does. The relatively smooth evolution from analog to digital was disrupted by the tAN and client-server computing explosion of the early- to mid-90s which left Westinghouse struggling to keep up with the surge in demand for wide area bandwidth. Client-server applications and files washed over into the wide area networks, often in ways that could not be anticipated. For example, new releases of software such as Microsoft Mail and cc:maii made zt much easzer for users to attach large "documents" (which could in fact be anything from a corpo- rate manual to a large b~nary image file) to a short e-mail message, leading to big increases in wide area communications wherever e-mail was already widely used--as it was at Westznghouse. New related applications like Lotus added to the "attachment explosion"; and an imminent decision to allow wider access by Westinghouse staff to the lnternet in late 1996 was expected to trigger yet another bandwidth explosion. Like others, Westinghouse zs making maximum use of compression techniques to try and get the most out of existing bandwidth. It com- presses data at a ratio of about 3 to 3.5:1, and runs voice at 8 Kbps (compared to 16 or 32 Kbps in most international public networks). At the same time, the bandwidth surge has been a key driver for it to examine alternatives, especially frame relay. On many routes, leas- ing more capacity means doubling capacity (eg from 64 Kbps to 128 Kbps) with a b~g hike in cost. The relative lack of granularity in h~gh- ly expensive international bandwidth (even worse higher up the scale where, e.g., users must usually leap from 2 to 34 Mbps) has forced users to look at alternatives. Westinghouse implemented its fi~t frame relay link (from Infonet) between its Pittsburgh head office and Brussels last year. On the voice side, the company may make greater use of virtual private networks (VPNs) Westinghouse is continually adding new sites to its corporate network. As well as the t 0 sites linked to the private network, Westinghouse has a further 10-12 international sites which must also be connected. With new digital network services, such as VPN, it’s much easier to route and consolidate traffic in ways that bring bigger discounts. More unusually, Westinghouse sells systems integration on its network to third parties, and has created a separate company, Westinghouse Communications, to sell services both internally and externally. In the U.S., Westinghouse already carries more external than internal traf- fic on its network, but in Europe ~t has only a couple of token contracts, basically extensions to domestic U.S. contracts. Westinghouse’s decision to sell services does not seem to affect its attitude to the question of outsouming bandwidth, where it expects to do more busi- ness. Rather, it wants to customize its offerings to compete with the big players. TeleGeography 1996/97 © TeleGeography, Inc. 1996 reliable. But for new users especially, frame relay is prov- ing hard to resist. Virtual nets: virtually here? On the voice side, the move to public solutions is much less pronounced. The market is only just being deregu- lated. Consequently, supplier offerings are not mature. Typically, these offerings do not take in many countries. Soundings taken by the Yankee Group Europe have shown a steady trend away from the PSTN, but no clear trend to VPNs. Yet it’s clear that virtual private networks or resold switched minutes--the distinction between the two is often hazy--are going to be the solution of choice for most corporations in the next few years. This is largely because of cost: the price-per-minute is falling rapidly under pressure from competition, shifting the crossover point and making it more difficult for corporations to jus- tify the use of private lines for voice. Initially, most inter- national VPN suppliers actually priced their offerings above the retail PSTN rate (there were few takers). Now, good negotiation will likely yield discounts of 40 percent on the PSTN, or even more. On the transatlantic route, big users often get a price of 15 cents a minute on VPNS between the U.K. and the U.S., against a published retail rate of 60 cents a minute. In contrast, the price per minute of compressed voice traffic on a fully loaded 2 Mbps circuit between the U.S. and the U.K. is in the region of $0.06 per minute. (This estimate is derived from the current 2 Mbps prices quot- ed by major suppliers, see Box 1.) To that, managers must add the cost of breakout at either end, which may add another 15 cents, and the cost of managing the cir- cuit. On the face of it, therefore, it may already be impossible to justify a 2 Mbps transatlantic circuit purely for carry- ing voice traffic. Since many corporations are already moving (or thinking about moving) data onto public ser~ vices, the economics of private voice networks become less compelling. On other routes, though, voice-over-IPL still makes a lot of sense for some companies, and especially where the networks already exist and have been very well-opti- mized. Suppliers report great difficulty matching the price-per-minute achieved in the best-run corporate voice networks, and often end up only winning business for peripheral sites. More importantly, perhaps, much private line voice is traveling on networks set up originally to carry data. Modern multiplexers can dynamically mute voice on the private network if the volume of data is low and, in this sense, the voice is said to be carried free--a somewhat misleading view, but enough to make an across-the- board move to VPNs more complex. Conclusion Despite the many factors driving corporations to put more of their international traffic on public network offer~ ings, the market among private corporations for IPLs is stubbornly buoyant. Inertia is one major reason: while international net new- bies go for managed offerings and existing net managers do the same when they are bringing up new routes, the old networks are there, are reliable, are often highly cost- effective, and there seems no good reason to tear them down. When demand increases, managers often find it easier to add another circuit, boosting the installed base of lines. And the capital equipment may not have been fully depreciated, making it harder to justify changing the financial structure of the corporate network. Private lines are also often still justified for the same rea- sons that they were justified 20 years ago: they are often cost-effective and very reliable. So long as that continues to be the case, corporations will continue to use them. As the great sea of bits flows back and forth between public, private and various intermediary networks, the one certainty is that no orthodoxy lasts forever; and in the febrile world of networking, forever means more than five years. This year, for instance, U.K. and U.S. authori- ties relaxed rules on the ownership of international cir- cuits between the two nations, making it possible for new operators to lay their own cables or to acquire IRUs as needed. In continental Europe, too, new legislation this year ended the TO monopoly over construction of infra- structure. The potential savings from building it yourself are vast: it costs $250,000 to buy outright a 2 Mbps half-circuit in a transatlantic cable. It costs at least as much to lease one for a year. You don’t need a degree in mathematics to see the opportunity. For the time being, public network solutions are clearly on the rise. Yet, if new services, technologies and own- ership options emerge, this paradigm could shift, and quickly. Remember, you read it here first. ¯ © TeleGeography, Inc. 1996 A Primer on International Simple Resale A PRIMER ON INTERNATIONAL SIMPLE RESALE The End Game Has Barely Begun by Gregory C. Staple International Simple Resale (ISR) is an obscure term even in the telecommunications industry. Yet, since 1992, when the U.S. and the U.K. first authorized ISR, it has become a powerful strategic weapon for carriers in a complex game for traffic and revenues. For example, using ISR circuits, a U.K. carrier, such as Mercury, can pick up and land traffic directly in the U.S., bypassing current correspondents, such as AT&T and MC[. But these U.S. carriers can retaliate in kind by using their own ISR facilities to enter the U.K. Thus, ISR may not immediately lead to more competition; major carders, each armed with ISR, may act cautiously, first launching ISR forays to probe their counterparts’ reac- tions. To understand the context for this new game, some background may be helpful. What is ISR? Broadly defined, ISR refers to the wholesale purchase of international private line (IPL) capacity from a facilities- based carder which is then resold to customers for switched telephone service. (An IPL is a leased circuit whose capacity is dedicated to the lessee.) Telephone service is typically provided by interconnecting the 1SR carrier’s circuits to the public switched telephone net- work (PSTN) at one or both ends of the international route. ISR thus may be defined simply as an IPL inter- connected to the PSTN at one or both ends. Customers can usually access an ISR carrier’s service on a dial-up basis. But if a customer has a substantial vol- ume of international traffic, it may hire a domestic leased line so as to connect directly to the ISR carrier’s nation- al switch. ISR carriers are often confused with companies reselling international switched services. Although some ISR car- riers may also be switched resellers, there are several important distinctions: ¯ An ISR carder pays a flat monthly rate for its transmis- sion capacity (i.e., the IPL). This rate typically reflects the cost of acquiring one-half of the 1PL from a carrier in Country A and a "matching" half circuit from a carrier in Country B. (Due to national licensing rules, few international carders own end-to-end "whole" circuits.) In contrast, a carrier reselling international switched services does not have its own dedicated capacity; it resells the switched service of a facilities-based international carrier, typically acquiring the service in bulk under a volume-sensitive discount. ¯ ISR carders do not pay settlement charges to the underlying facilities-based carders. Traffic handled by ISR carders therefore bypasses the international accounting rate regime. (For an explanation of this regime, see Box 1. Current accounting rates on major routes are listed in the tables following this essay.) In contrast, the interna- tional traffic handled by switched service resellers, such as call-back companies, is subject to international settle- ment payments because this traffic is bundled with the switched traffic of the underlying carrier for foreign ter- mination. ¯ ISR carriers are able both to originate and terminate international traffic because they can interconnect their IPLs at both ends. By comparison, switched resellers can only originate service, although some switched resellers may operate in more than one country and thus appear to offer two-way services. The Economics of ISR ISR is economically attractive to two types of carders: ( 1 ) entrepreneurs which cannot acquire their own interna- tional facilities for regulatory or economic reasons (e.g., if the government will not issue additional facilities-based international licenses); and (2) incumbent facilities- based carriers, especially those with a payment deficit to major correspondents (i.e., carriers which send out more traffic than they receive) (see Table 1). For entrepreneurial carders, ISR is primarily an arbitrage business. The profitability depends upon the ISR card- er’s ability to exploit the differential between the current retail prices for international telephony and the price of the underlying components--that is, the IPL circuits plus the domestic access charges. The larger the price spread, the greater the ISR carrier’s potential profits (see Box 2). Regulation of IPL prices and domestic access charges are thus of prime importance to ISR carriers who live in con- stant fear of a "price squeeze." Gregory C. Staple is the Editor of TeleGeography 1996/97 and a Partner in the Washington, D.C. communications law firm of Koteen & Naftalin, L.L.P. The economics of ISR are not so clear cut for an incumbent facili- ties-based carrier ISR provides such carriers an opportunity for reducing foreign settlement pay- TeleGeography 1996/97 © TeleGeography, Inc. 1996 Box 1. Accounting Rates and How they Work Accounting rates provide a common method of compensating originating and terminating carders for carrying telephone calls over both networks. The system is largely transparent to users and generally works as follows: 1. International carriers negotiate accounting rates on a mute-by-mute basis. A rate is agreed per minute for landing traffic in either direction based on the sum of both carriers’ costs, although the cost- linkage is often quite loose. The rate is commonly stated in U.S. dollars or Special Drawing Rights (SDRs), a monetary unit whose value reflects a "basket" of major currencies. 2. On any given route, one carrier pays settlements to another carrier only to the extent that there is a traffic imbalance---that is, one carder has terminated a greater volume of telephone minutes than the other carrier. The originating and terminating carrier usually divide the accounting rate 50/50 to determine the per minute settlement rate. 3. A carrier’s net revenue for international service is a function of the accounting rate as well as the collection charge (see Figure 1). If traffic is balanced, the value of the accounting rate is essentially irrelevant since no settlement is necessary and each carder’s revenue will depend directly on its collection charge. 4. Where traffic is imbalanced, the accounting rate may have a significant effect on the commercial options of the two carriers. If a carrier has a significant traffic deficit, the settlement payments which it must make to its foreign correspondent limit its ability to reduce its collection charges. Conversely, a carrier with a net traffic surplus has little incentive to operate more efficiently or to reduce the accounting rate because of the net settlement benefits it receives under the status quo. 5. Carriers which have relatively lower collection charges (often due to the competition from other carriersI and a net traffic deficit are dissatisfied with the current accounting rate regime: it tends to subsidize high cost monopoly carriers at the expense of lower cost carriers and end-users from competitive regimes. That is also why carriers with a net traffic outflow tend to favor ISR. If ISR is legal, such carriers can land their traffic overseas without paying foreign settlement charges. Some carriers which dislike the status quo also want the right to establish their own foreign affiliates so that they can provide end-to-end service and "settle" with themselves. Figure 1. Accounting Rate Monetary Flows The Anatomy of an International Phone Call How International Revenues are Shared Accounting Rote ( Internal price between PTOs for a jointly-pro- vided service) Collection Charge (The amount charged to the customer by the PTO) Settlement Rate ( Payment from one PTO to another, normally haft the accounting rate) I Country A sends 1001 Country B sends 150 m~nutes to Country B minutes to Country A b= ed at1 umt pertain b= ed at 75 umt pertain. Country A Country B ¯ Collects lOO umts ¯ Collects 112 5 umts ¯ Pays 25 umts to Country B ¯ Pays 37 5 umts to Country A ¯ Recezvas 375 umts from B ° Recewes 25 umts from B Note: This box adapted from Oirecfion of Traffic 1996© ITUFFGI © TeleGeography, Inc. 1996 A Primer on International Simple Resale Box 2. Sample Break-Even Calculation for ISR Carrier IPL IPL Cost IPL Cost IPL Switch Term. ISR T~l~globe ISR Route Capacity Origin. End1 Term. End1 Cost]Min.2 Cost/Min.3 Cost/Min.4 Tariff/Min.5 Canada- E-1 circuit $35,000 $42,699 $0.052 $0.023 $0.075 $0.24 U.K. (30x64 Kbps) Canada- 766 Kbps $42,800 $52,374 $’0.157 $0.044 $0.201 $0.42 Austraha Note: All figures m Canad=an dollars (as of September 1998, C$1 00=US$0.73). 1. Prices before d=scounts. 2. Based on 50,000 minutes per 64Kbps c=rcuit per month and assumes 5 voice paths per 64Kbps circuits 3. U.K. and Australian swztched termznat=ng costJmmute figures are based on BT and Telstra tariffs converted to CanaOan dollars and an estimated distribut=on of traff=c by tzme of day and destinatzon, 4. Local access and contribution costs are not zncluded as they are common to both ISR and GAT scenarios 5. The T~l~globe Global Access Tariff (GAT) provides a wholesale swztched rate for large customers and resellers. Tariff here is based on a 13AT rate of $0 217/mznute; a surcharge of $0.20/ca11, and a dzvers=ty discount of 1% An average call duration of 7.5 m=nutes/call was used. Source: Bell Canada, T~l~globe © Tele6eography, Inc. 1996 ments but may provoke carriers’ foreign correspondents as well as its domestic competitors. Thus, for incumbent carriers, ISR may be characterized as a non-cooperative game where the welfare of each carrier is linked; infor- mation about other player’s preferences is limited; and hence the optimal strategy and payoffs are not easily cal- culated in advance.* This has made most carriers care- ful in playing the ISR game, with many preferring merely to mirror the moves of their principal competitors (i.e., to adopt a "tit-for-tat" strategy). Learning the Game A numerical example may clarify the reasons why most incumbents have been reluctant to take a more aggres- sive tack towards ISR. Assume, for instance, that in 1995 Incumbent # 1 in Country A sent 1 O0 million min- utes to Incumbent #2 in Country B and received back only 75 million minutes, generating a net deficit of 25 million minutes or $12.5 million (i.e, the accounting rate was $1 and the settlement rate was $O.50). Assume also that Country A and B both permit ISR, and that there is at least one other incumbent in each country. In 1996, if Incumbent #1 decides to send 25 million minutes to Country B over ISR facilities, thus bypassing the accounting rate system and clearing its deficit, Incumbent #2 might respond by doing likewise, in which case the deficit would remain the same. Alternatively, Incumbent #2 might retaliate by shifting a greater volume of outbound traffic to ISR facilities; rout- ing traffic to a competing carrier in Country A; or by establishing (or buying) its own affiliate in Country A so as to "self correspond." To make things even more corn- For a general introduction to cooperative and non-cooperative games, see H. Scott Bierman and Luis Fernandez, Game Theory With Economic Appfication (Addison-Wesley, Reading, MA, 1993]. plex, Incumbent #2 also could seek to divert Country A~s inbound traffic from Countries C, D and E via its own ISR facilities, thus depriving Country A of inbound settlement payments from third countries and worsening its current deficit situation. (There are non-traffic related counter- moves that many carriers could initiate, of course.) Table 1. Who Cares Most About ISR? Estimated Balance of Settlements for Public Switched Services (1995) Traffic in millions of minutes and settlements in millions of US$ Traffic Surplus (Deficit) Settlement Country Surplus (Deficit) Argentina 120 85 Australia n.a. (25) Brazil 176 135 China n.a. 480 Germany (1266)* (801) Hong Kong (93) (78) India 464 254 Japan (384) (! 51) Korea 115 110 Mexico 1164 444 Philippines 517 235 Poland 268 111 Portugal 241 148 Russia 161 183 Saudi Arabia n.a. (135) Singapore n.a. (41) Sweden n.a. (40) Switzerland (339) (135) Turkey 331 173 U.K. 5 (159) U.S. (8623) (4500) *Data for 1994 Note: Settlements based on estimates ~n Direction of Traffic 1998 © ITU/TGI. TeleGeography 1996/97 © TeleGeography, Inc. 1996 Faced with such scenarios, most countries have adopted some basic legal ground rules for ISR to make the game more manageable. These rules typically have been designed to reduce the risk for incumbent carriers subject to potential competition from foreign-affiliated ISR providers. The Law of ISR The legal regimes for ISR fall into two main groups: "pro- tectionist" and laissez faire. (see Box 3). Protectionist countries typically require prior authorization for 1SR car- riers or permit ISR service only on those routes where "equivalent" opportunities exist at the foreign end. This group of countries currently includes the U.S. and Canada. The free market group, which includes Australia, New Zealand, Sweden and the U.K., permits ISR carriers to enter the market upon completing a simple notifica- tion process (the U.K. requires a license) and allow ser- vice on any route even though similar opportunities may not exist at the foreign end (i.e., a company may connect an IPL with the public network at one end only, known as "one-ended ISR"). Some countries, notably Denmark and Finland, fall somewhere in between these two regimes. Hubbing rules for ISR also distinguish the protectionist and laissez faire camps, although the line is more blurred. As used here, hubbing refers to the indirect routing of switched traffic from Country A to Country B using ISR facilities connecting Country A to a third coun- try where traffic is landed and then refiled to Country B either via the PSTN or another ISR link. Box 3. Legal Rules for ISR in Selected Countries Country i Entry Foreign Ownership Hubbing Australia No hm~ts. Canada Sweden Requires enrollment with Australian Telecommumcat~ons Authority (Austel) under the International Service Providers Class License (ISPCL). One-ended and two-ended ISR permitted without "equivalence" test but Austel retains power to intervene where a foreign-affihated ISR carrier is misusing ~ts market power or offshore regulatory status to substantially lessen competition. Letter notification required to Canadian Radio-television Commission (CRTC) ISR permitted only to countries prowding "equivalent" opportumt~es (i.e., only two-ended ISR is lawful) but CRTC does not review equivalence. Faciht~es-based local exchange carriers (i.e., Bell Canada and other Stentor companies) may not provide ISR but pohcy ~s under review. Open entry. L~cense only required from the National Post and Telecom Agency (NPTA) if resale activity is of a "considerable extent" There ~s no "equivalency" requirement, i.e., one-ended and two ended ISR is lawful. Requires grant of Pubhc Telecommunications Operator (PTO) hcense by Department of Trade and Industry (DTI). Prior "equivalency" rules lifted in June 1998 so that ISR is now permissible on any route, but government retains power to prevent abuses by foreign-affd~ated ISR carriers. Where foreign carrier has a monopoly, proportionate return is required for ISR and non-ISR traffic; parallel accounting rates are also required on such routes. Requires prior authorization of Federal Commumcat~ons Commission (FCC) under Section 214 of the Commumcations Act. Authority granted route-by-route to countries where Effectwe Competitive Opportumties (ECO)--Le., substantially simdar opportumt~es--exist. Thus, one-ended and two-ended ISR only permitted to countries meeting ECO test. No limits. No Jlmlts. No hm~ts. No hm~ts, but foreign affd~ated ISR carriers must show that they lack market power to serve home country. No restrictions but subject to Austel oversight. Only with the consent of all concerned part~es (i.e., hub and destination countries). Pohcy under review. Unhm~ted No limits but government retains power to intervene to prevent discriminating practices. Switched hubbing only, i.e., traffic from U.S. to Country B may be routed via ISR famht~es in Country C provided it is refiled to Country B via Public Switched Telephone Network (PSTN) in Country C and not via an IPL. Note: At September 1996, ISR was also lawful, in principle, m Denmark, F~nland, Germany, the Netherlands and New Zealand, subJect to local hcensing requirements © TeleGeography, Inc. 1996 © TeleGeography, Inc. 1996 A Primer on International Simple Resale F The Impact of ISR: Carriers, Traffic and Accounting Rates 1995 Total Telephone Accounting Rate Per Minute (SDR) Outbound ISR Traffic (million minutes) 40-80 Number of ISR Carriers (6/96) 50 20 -10 40 50 Country Route Before (1991) After (1995) Australia AU-U.K. 0.70 0.39 AU-U.S.* 0.68 0.308** Canada 250-300 CN-U.K. ~0.40 0.20 Sweden 20-30 SW-U.K. 0.40 0.23 SW-U.S. 0.50 0.25 U.K. ~250 U.K.-U.S. 0.53 0.25 (1996) U.S. 200-250 U.S.- C N 0.20 0.15 U.S. ISR apphcations on th~s route were still pending at September 1996 ** This =s a 1996 average for the route reflect=ng different Austrahan rates for term=nating traffic in urban, rural or mob=le networks. Note: Accounting rates are for the largest carrier on each route (AT&T or BT). 1 SDR --$1.4585 at Sept. 1, 1996. © TeleGeography, Inc. 1996 If unlimited hubbing is permitted, an ISR carrier may be able largely to supplant the role of existing facilities- based carriers in a given country even on routes where ISR is not permitted. This can be done by collecting out- bound or inbound traffic to or from closed markets at one or more off-shore "hubs" and then linking the hubs to the target country by ISR facilities. Further, if there are no limits on foreign ownership of the ISR carrier, such hubbing arrangements may lead to the de facto market entry by facilities-based foreign carriers. Canada is instructive in this regard. In Canada, ISR car- riers need not obtain route-by route approval although, in principle, ISR is permitted only where "equivalent" opportunities exist at the foreign end. This lax regulato- ry regime has led some domestic long distance carriers, such as Unitel and Sprint Canada, to use ISR facilities to hub or "refile" Canada-overseas traffic to third countries via the facilities of affiliated U.S. carriers (AT&T owns 33 percent of Unitel, recently renamed AT&T Canada Long Distance; Sprint owns 25 percent of Sprint Canada). Inbound Canadian traffic is said to be hubbed in reverse fashion, in some cases with the aid of off-shore foreign Box 5. How do U.S. Carriers Use Their International Facilities? Total Private Line, Public Switched and Idle 64 Kbps Circuits by Region, December 1995 Private Public Switched Total Idle Total Routes Line Network In Use Circuits Available North America 7,196 67,588 74,784 2,736 77,520 C. & S. America 1,776 6,955 8,731 6,274 15,005 Caribbean 517 5,349 5,866 1,779 7,645 Western Europe 9,997 22,389 32,386 54,593 86,979 Eastern Europe 241 2,886 3,127 1,470 4,597 Middle East 506 2,560 3,066 266 3,332 Africa 199 2,051 2,250 181 2,431 Asia 5,067 13,185 18,252 26,605 44,857 Oceania 998 3,125 4,123 24,439 28,562 Other 0 60 60 N/A 60 Total 26,497 126,148 152,645 118,343 270,988 Note: Data based on FCC czrcuzt status reports filed by U S, carriers and are for AT&T, MCI, Sprint and WorldCom only Data are for circu=ts originating =n continental U.S "Idle" circuits are c=rcu=ts owned by a carrier atyear end but not in use Totals are for all c=rcuzts to all countries w=thin a reg=on. Limzted idle circuits to N American routes reflect absence of cable famhtzes. By comparison, large =dle capacity on routes to Asza and Oceama reflect recent introduction of new cable facil=- ties Satelbte capamty utd~zat~on ~s generally not reflected by this data because U S carriers do not acqu=re international satelbte capac=ty m advance © TeleGeography, Inc. 1996 TeleGeography 1996/97 © TeleGeography, Inc. 1996 affiliates (Sprint and AT&T both have ISR authority in the U.K.). Critics of unrestricted ISR allege these hubbing activities are contrary to Canada’s best interests: they undercut the revenues of T~l~globe, Canada’s sole facilities-based overseas carrier. Just as importantly, this type of cross- border hubbing services tends to make Canada increas- ingly dependent on the U.S. because, as more and more of Canada’s overseas traffic is rerouted via the U.S., any Canadian carrier (T~l~globe’s monopoly expires in 1997) will find it increasingly difficult to negotiate competitive foreign operating agreements in its own right. Ironically, analogous concerns have been expressed by U.S. carriers. But, thus far, they have had a somewhat more sympathetic regulator Under FCC rules, U.S. carri- ers may not indiscriminately refile traffic to third coun- tries. "Switched hubbing" is permitted, but this limits ISR carriers to routing third country traffic via a country which provides equivalent ISR opportunities and then refiling the traffic through the PSTN; traffic may not be refiled via another IPL. The potential for ISR carriers to engage in switched hub- bing and the difficulty of policing such arrangements is also a crucial concern of trade negotiators seeking to open national markets for basic telecommunication ser- vices on non-discriminatory (i.e., reciprocal) terms. U.S. and other World Trade Organization (WTO) parties assert that a country which does not agree to open its market might still benefit from an agreement if any party there- to (such as the U.K.) permits one-ended ISR. In that case, a non-party could establish a U.K. based ISR carri- er which could pick-up and deliver traffic throughout the liberalized block of countries as well as its home country. Box 6. Call-back = Cash-back: Why Call-back May Help Low Income Countries It is often alleged that telephone call-back services, which permit local callers to obtain a lower priced international dialtone from a foreign carrier, deprive the local Public Telecommunications Operator (PTO) of much needed inter- national revenues. But while call-back services may lead to reduced revenues for the national PTO, they do not nec- essarily penalize the developing country as a whole. Consider the following case where a resident of a developing country spends US$10 on making a call direct via a national PTO (at US$2 per minute for a 5 minute call) and another spends the same amount with a call-back com- pany (at US$ I. 33 per minute, a 7.5 minute call). Also assume that the accounting rate with the distant country is US$1.20 per minute, making the settlement payment US$O. 60 per minute. (See Box I for a review of the account- ing rate system.) In the first case, the settlement payment out of the country is US$3,00 (5 minutes x $0.60). In the second case, the settlement payment into the country is US$4.50 (7.5 minutes x $0.60). The developing country PTO loses $ 7. O0 in local currency for the call but gains $4.50 in hard currency settlement payments. Furthermore, in the call- back case, there are no costs incurred for the local PTO in billing for the service or in debt collection (done by the call-back provider). For the country as a whole, however, the main advantage is that the call-back user was able to make 50 percent more calls for the same price. For many developing countries, settlement payments are now a major source of telecommunications revenue. In sev- eral Latin American and Caribbean countries, net settlement payments from the U.S. alone already make up more than half of all telecommunications revenue. In Africa too, where call-back is popular due to the high price of inter- national calls, net inward settlement payments from the United States have tripled in just five years. A high per- centage of this windfall has been poured back into network development. Of all the different "alternative calling procedures, call-back is certainly the most suited to the needs of developing countries. Some alternative calling procedures stimulate traffic without bringing financial benefits to the local PTO (this is the case, for instance, with Internet telephony, International Simple Resale and other PSTN-bypass systems). Others bring financial benefits to the local PSTN (via the settlement payment mechanism) but do not necessarily stimulate traffic by offering lower prices (for instance, calling card traffic, country-direct services). Only call-back performs the trick of both stimulating traffic and raising cash for the local PTO. Ada pted from Direction of Traffic 1996 © ITU/-I’B I J © TeleGeography, Inc, 1996 A Primer on International Simple Resale The Impact of ISR Military strategists well know that a new weapon fre- quently need not be launched in order to be effective. ISR provides a similar deterrent. In l gg2, many observers predicted that ISR would lead to wholesale accounting rate bypass and price cutting on major trans- Atlantic routes especially given the large amount of unused transmission capacity on key routes. But even though raw (IPL) capacity remains abundant, the volume of global ISR traffic remains modest (see Box 5). To date, with minor exception, no major carrier has wished to launch a "first strike." For example, carrier traffic reports filed with the FCC show that as of July 1996, ISR traffic was only significant on the U.S.-Canada route (with approximately 400 mil- lion minutes of twoMway traffic in 1995). On the U.S.~ U.K. route, no major carrier had more than a small vol~ ume of traffic and many had not even begun ISR service (see Box 4). (By comparison, U.S. resellers of interna~ tional switched services carried over 2.3 billion minutes in 1995, at least one third of which is estimated to be "callMback" traffic. The impact of this "call-back" traffic on less developed countries is discussed further in Box 6.) Outside the U.S., the story is somewhat different. ISR carriers now carry over 5 percent of the U.K.’s outbound traffic. This is largely because the U.K. has long permit- ted one-ended ISR, thus permitting ISR carriers to hub traffic from non-lSR countries through the U.K. and onward to North America or Asia (via Australia) using their own dedicated IPL network. In !995, ISR carriers in the U.K. collectively carried over 250 million minutes of international traffic. Similarly, in Australia, ISR carri- ers, including companies affiliated with foreign carriers-- such as Singte!, Sprint and BT--now account for at least five percent of the outbound market in terms of minutes. The Future of ISR So long as the international accounting rate system leads carriers from competitive regimes, such as the U.S., Canada and Sweden, to run significant deficits with car- tiers operating in more restrictive markets, ISR is likely to have continuing appeal. (Even where a carrier does not have a deficit, ISR may still provide a means to reduce net foreign settlements.) The greater the deficit, the greater the incentive to use ISR to bypass high foreign settlement rates (see Table 1). As we have seen, howew er, the risks of pursuing such a strategy are consider- able--others can play the same game. The uncertain benefits of ISR have led several countries-- notably, Chile, the Philippines and Mexico--to encourage competition solely by granting multiple facilities-based international licenses. Each of these countries have a sig- nificant traffic and revenue surplus--a surplus which might be eroded by foreign carriers if ISR is permitted. But so long as ISR is banned, the net in-flow of foreign settlements may be used to help jump-start multiple new carriers. ISR also has mixed support in the European Union (EU). Apart from the Scandinavian countries and the U.K., other EU members have decided to liberalize the market primarily by granting additional facilities-based licenses for international service. From January 1998, ISR will be permitted in most EU states, but as competition drives down accounting rates on key routes (see Table 2 follow- ing this essay) and as traffic becomes more balanced, the incentive to use ISR on intra-European routes may be rel- atively limited. It would be a mistake, however, to suggest that the ISR game is almost over. Apart from the numerous routes where large traffic and revenue imbalances remain, ISR is also likely to be a major weapon for the new generation of global alliances. Two of these alliances, Concert and Global One, have put in place sophisticated internation- al backbone networks. Once ISR becomes lawful in more countries, these new backbone networks may wel! become the primary means for some of the world’s largest carriers to route traffic to and from their national affiliate in a way which provides the optimal mix of col- lection charges and settlements for their owners. When that begins to happen, the ISR game may finally begin in earnest. ¯ TeleGeography 1996/97 © TeleGeography, Inc. 1996 International Accounting Rates Table 1. United Kingdom Accounting Rates for OECD Countries, 1991-1996 Note: Percent Country 1991 (US$) 1996 (US$) Change Australia 1.00 0,56 -44% Austria 0.58 0.50 -15% Belgium 0.47 0.43 -7% Canada n.a. 0.29 n.a. Denmark 0.53 0.35 -34% Finland 0.62 0.37 -40% France 0.47 0.37 -22% Germany 0.49 0.22 -55% Greece 0,68 0.62 -10% Iceland 1.03 0.61 -41% Ireland n.a. 0.28 n.a. Ira ly 0.58 0.57 -3% Japan 2,00 1.46 -27% Luxembourg 0.46 0.41 -10% Mexico 2.10 1.75 -17% Netherlands 0.48 0.43 -11% New Zealand n.a, 0.88 n.a. Norway 0.56 0.35 -38% Portugal 0.72 0.61 -16% Spain 0.59 0.62 4% Sweden 0.57 0.34 -42% Switzerland 0.51 0.30 -41% Turkey 0.86 0.57 -34% US BT 0.83 0.36 -56% US Mercury 0.76 0.44 -42% Figures based on SDR convers=ons of 1 SDR=$1.43 for average 1991 and 1 SDR-~1.46 for average 1996 as (of September 1). Source" Oftel Inc.~ © TeleGeography, 1996 ounting Rate Trends for OECD Countries, 1991-1996 To Europe (SDRs) To Asia-Pacific (SDRs) Country 1991 1994 1995 1996 1991 1994 1995 1996 Belgium 0.41 -- -- -- 1.23 -- -- -- Denmark 0.47 0.40 0.30 -- 2.85 2.34 1.61 -- France 0.42 0.33 -- 0.31 1.20 0.80 -- 0.68 Greece 0.53 0.42 0.42 -- 2.18 1.22 1.07 -- Ireland 0.53 0.48 -- -- 2.15 1.71 -- -- Italy 0.37 0,34 0,34 0.30 1.53 0.97 0.92 0.56 Netherlands 0.42 0.30 -- -- 1.45 0.95 -- -- Norway 0.45 -- -- -- 1.25 -- -- -- Spain 0.41 -- -- -- 1.59 -- -- -- Sweden 0.40 -- -- -- 1.22 -- -- -- Japan 1.71 1.15 -- 1.09 1.34 -- New Zealand 1.45 1.14 -- -- 0.97 -- Note: All figures are averages in SDRs for each regzon Source: OECD 0.75 0.70 0.60 J © TeleGeography, Inc. 1996 © TeleGeography, Inc. 1996 International Accounting Rates United States Accounting Rates, 1992-1996 (US$) Table 3. Country Argentina Australia (Telstra) Austria Brazil Canada (Stentor) Chile (ENTEL) China Columbia Croatia Cyprus Czech Republic Denmark Finland (Telecom Finland) France Germany Greece Hong Kong Hungary India Indonesia Iran Ireland Israel Italy Japan (KDD) Korea (Korea Telecom) Kuwait Luxembourg Macau Malaysia (TM) Mexico Morocco Netherlands New Zealand (TNZI) Norway Philippines (PLDT) Poland Portugal Russia (Rostelcom) Saudi Arabia Singapore Slovenia South Africa Spain Sweden (Telia AB) Switzerland Taiwan Thailand Turkey United Arab Emirates United Kingdom (BT) Venezuela Yugoslavia 1992 1993 1994 1995 1996 1.65 1.65/1.54 1.52 1.43 1.43 0.83 0.76 0.59 0.59 0.58 1.10 0.83 0.89 0.67 0.51 1.50 1.40 1.27 1.14 1.12 .28/.24 .28/.24 .26/.22 .24/.20 .22/. 14 1.60/1.15 1.60/1.00 1.10 1.10 1.00 3.37 2.93 2.91 2.67 2.14 1.55 1.50 1.40 1.30 1.27 1.71 1.38 1.34 1.04 0.73 2.28/1.65 1.66 1.49 1.41 1.39 n.a. 1.24 1.19 1.04 0.88 1.38 1.38 1.19 0.74 0.58 0.89 0.90 0.74 0.59 0.58 0.96 0.97 0.62 0.54 0.35 1.10 0.83 0.51 0.39 0.23 1.65 1.55 1.41 1.26 1,09 1.60 1.20 1.00 1.00 1.00 1.40 1.24 1.34 1.34 0.73 2,00 1.90 1.80 1.80 1.80 1.80 1.80 1.80 1.58 1.40 3.00/2.50 3.00/2.50 3.00/2.50 3.00/.2.50 3.00/2.50 1.10/.92 0.93 0.82 0.67 0.44 2.28/1.9/1.63 2.16/1.85/1.40 2.16/1.85/1.40 1.90/1.63/1.23 1.18 1.65 1.5/1.10 1.22/.82 0.71 0.53 1.31 1.04 0.94 0.94 0.92 1.60 1.44 1.41 1.26 1.24 1.58 1.59 1.71 1.71 1.68 1.38 0.97 1.04 0.74 0.58 1.80 1.80 1.65 1.50 1.35 1.80/1.20 1.15 1.05 1.00 0.89 1.10 1.00 0.91 0.67 0.67 3.03 3,04 3.27 1.78 1.75 0.69 0.69 0.59 0.37 0.36 1.65 0.83 0.89 0.59 0.44 1,10 0.97 0.74 0.45 0.44 1.68/1.25 1.68/1.25 1.34 1.23 1.20 1.30 1.25 1.20 1.15 0.95 1.65/.69 1.49/.89 1,41/.74 1.20/,74 .95/.73 2.60 2.60 2.60 2.60 2.12 2.20 2.20 2.20 2.20 2.20 0.85 0.86 0.92 0.92 0.90 1.71 1.71 1.49 1.11 0.73 1.80 1.50 1,20 1.20 1.00 2.06/1.38 1.80/1.10 1.78/1.04 1.44/.95 0.64 0.69 0.69 0.37 0.37 0.18 1.11 0.84 0.91 0.52 0.51 1.40 1.20 1,20 1.20 1.20 !.75 1.60 1.60 1.55 1.50 1.93 1.66 1.78 1.63 1.24 2.00/1.30 2.00/1.30 2.00/1.30 2.00/1.30 2.00/1.30 .74/.52 .81/.48 0.49 0.37 .37/.22 1.30 !.30 1.30/1.00 1.30/1.00 1.15/1.00 1.73 1.49 1.34 1.19 1.17 Note: Rates are for ]argest career serving the route, different accounting rates may apply to competing carriers, Where two rates are shown, there are peak/off-peak rates or growth-based rates (Le., traffzc above a benchmark level =s ehg~ble for a lower rate). 1996 data currentto September 1. All figures expressed m US$. Source: FCC © TeleGeography, Inc, 1996 TeleGeography 1996/97 © TeleGeography, Inc. 1996 SETTLEMENTS FOR PHONE SEX Major Providers of International Audiotex Estimated Incoming Audiotex Traffic, 1995 Country Million MiTF per month 1. Netherlands Antilles 9.0 2. Guyana 9.0 3. Domimcan Republic 8.0 4. Hong Kong 8.0 5. S~o Tome 4.2 6. Niue 4.2 7. Canada 4.0 8, Philippines 4.0 9. Austraha 3.4 10. Israel 3.0 11. Chile 2.3 12. Northern Marianas 2.3 13. USA 2.3 14. Norfolk Islands 2.0 15. Moldova 1,7 16. St-Pierre and Miquelon 1.1 17. Portugal 1.1 18. Hait~ 1.0 19. Jamaica 1.0 20. Antigua 1.0 Other 3.8 TOTAL 76.4 Source’ Industry interviews. Global Audiotex Traffic Growth, 1992-1996 Estimated International Traffic (MITT) 1200 1000 2OO 225 1050 95O 1992 1994 1996 ©TeleGeography, Inc. 1996 W ~hat do you get when you cross the internation- al accounting rate system with the demand for phone sex? The answer: a $2 billion industry that is responsible for more than 900 million minutes of international telephone traffic. Since 1990, stricter domestic regulation of "adult" content provided over public telephone lines has pushed a piece of the phone sex industry across international boundaries. By relocating a sex-talk business in a relatively distant country, service providers can take advantage of high international call prices and the telephone company billing system while bypassing local regulations. The consumer in the originating country (e.g., the U.S.) places a call to an advertised number in another country (e.g., Guyana). The consumer then pays for the call on the next phone bill, like any other per-minute long distance charge. The only difference from a normal international call is that the settlement payment made by the originating (billing) carrier to the telco in the terminating country is split with the service provider. Therefore, the countries with the highest settlement rates offer the most lucrative environ- ment for international audiotex. Not all audiotex is pornographic. The types of services available range from sports scores to live chat lines to pre- recorded fantasies. And in many cases the terminating country may not even be the host for the service. In this scenario, the terminating country acts as a collection vehi- cle for inward traffic which is then rerouted (by private line) to an audiotex provider in a third country or even back to the country from which the incoming calls originate. Settlement Rates for Audiotex Providers (All figures in US$) Country U.S. W. Europe Country Code (9/96) (1995 est.) Australia 61 0.29 0.45 Canada 1-604/416 0.11 0.30 Chile 56 0.50 0.85 Dominican Republic 1-809 0.55 0.70 Buyana 592 0.85 0.80 Hong Kong 852 0.50 0.75 Israel 972 0.59 9.90 Moldova 373 1.04 0.60 Netherlands Antilles 597 0.38 0.70 Niue 683 1.50 1.75 Northern Marianas 670 0.30 0.85 Philippines 63 0.60 0.75 Portugal 351 0.48 0.40 Sao Tome 239 1.17 1.00 Note: The settlement rate is the per-minute payment made by the orig=natmg carrier to the terminating career. The payment ~s ~p~cally spht with the ser- wce prowder. U S, rates for peak only. W. Europe are aggregate est=mates. © TeleBeography, Inc. 1996 The internet TeleGeography 1996/97 © TeleGeography, Inc. 1996 © TeleCeography, Inc. 1996 Internet Traffic Statistics: An Overview INTERNET TRAFFIC STATISTICS: AN OVERVIEW n a perfect world (one crafted by geographers) the Internet would map itself. And the billions of bits flow- ing from one computer to another would be counted too. The technology already exists--computers con- nected to the lnternet would simply use the Internet Protocol (IP) to add a header encoding the geographic origin of each data packet as well as its destination. This scheme is not really far-fetched. Geographic tags can already be added to World Wide Web pages. And tomor- row’s computers may well include a standard radio module which, like a cellular phone, constantly updates the com- puter’s position via the network of Global Positioning Satellites or a terrestrial system. Right now, however, map- ping the Internet’s ever changing traffic patterns still pre- sents a colossal challenge. How much traffic flows between Internet Service Providers [ISPs) in the United States and other countries? What is the mix of traffic? How much is e-mail, how much is generated by the World Wide Web, and how much is real-time traffic, such as Internet telephony or video conferencing? And how has the changing number and distribution of Internet com- puters (hosts) affected the volume and balance of traffic between different countries and backbone networks? in TeleGeography 1995, we provided rudimentary answers to some of these questions based upon a late 1994 survey of traffic flowing over the NSFNet, then underwritten by the U.S. National Science Foundation (NSF) (see Box 1 below). While not all lnternet traffic flowed over NSFNet, the bulk of it did. But NSFNet was shut down beginning in November 1994, and there is no longer a central Internet backbone. The traf- fic data now available thus reflects only isolated sections of the network, lnternet-wide measurements simply do not exist. Performance Measurements The business press has recently discovered the lnternet sta- tistics gap as stories about network congestion begin to make front-page news (see Box 2, "Mr. Quarterman’s Internet Weather Report," on the next page). During peak hours, many Internet sites are hard to reach ("connection refused"); data packets are lost (20 percent or more at some network exchange points) and downloading files can be maddeningly slow. Some Internet engineers also expect the Net to suffer from periodic service "brownouts" or worse as operators scramble to meet the demand for new band- width-hungry services, such as real time Internet audio and video communications. In this environment the demand for statistics has gathered force--statistics which will help ISPs and backbone opera- tots identify the "trouble spots," test the impact of new Box 1. The Traffic Statistics Dry Up Traffic on the U.S. National Science Foundation (NSF) Backbone, by category of service December 1994 begmmn9 of trans~on to new NSF arch~ecture Until its decommissioning on April 30, 1995, the NSF Backbone Network was the most heavily used, large-scale Internet interconnection facil- ity in the world. It was also a superb window on Internet usage, both for total volumes and breakdowns by type of traffic. Under the new architecture, traffic flows on commercial networks, such as ANS, MCInet, and Sprintlink, and it is no longer possible to measure Internet use by protocol as before. TCP/UDP are basic Internet standards that allow up to 128,000 different network applica- tions and services to be provided over the Internet. Non-TCP/UDP standards include Open Systems Interconnection (OSI) and various spe- cial net~vork services. Source: Merit, ftp//mc.ment-edu/nsfnet/stat~st~cs/ I~ Nen-TCP/UDP fa Other TCP/UOP (including gopher, [] Name Lookup [:3 Interactive [3 Mad (mcluthng USENET) [] File exchange 188QQ 14000 ©TeleGeography, Inc. 1996 TeleGeography 1996/97 © TeleGeographg, Inc. 1996 Box 2. Mr. Quarterman’s Internet Weather Report John Quarterman, a computer consultant in Austin, Texas, has been mapping the furthest reaches of the Internet since the 1980s (see "The Ends Of The Matrix" in TeleGeography 1993). And while others have made headlines with predictions about the lnternet’s imminent demise, Quarterman’s research group, the Matrix Information and Directory Service (MIDS), has been quietly compiling data on the Net’s actual performance. Since 1994, MIDS has measured the round-trip time for sending a message from its Austin computer to approx- imately 4,500 Internet sites world wide. MIDS uses a program known as ping (technically ICPM ECHO) to veri- fy the presence of hosts computers. The MIDS test sites are "pinged" five times per site every four hours, seven days a week. The record of each ping is mapped and each day’s maps are then animated using the MPEG movie protocol. Soon after, the results are posted on the World Wide Web (http://www.mids.org/mids/weather.html). These Internet Weather Reports (IWRs) show plenty of storms. But Quarterman suggests that longer average response times for any particular site are often due to local causes, such as an overloaded computer, a slow drive or a congested ISP which don’t really effect overall Internet performance. Still, IWRs do show an interesting cycli- cal feature: Congestion increases during the work week with five days of relatively high latencies (round trip times) followed by two days of shorter latencies. Holiday seasons usually produce better performance but dur- ing the last Christmas-New Year season, latencies went up, says Ouarterman, perhaps because lots of people "got Internet connections for Christmas." The IWRs also show that from January 1994 to January 1996 there was a 30 percent improvement in mean latencies. That is, during this period, the average round trip time went down, which means that the quality of the Internet actually got better--not worse. Quarterman thinks that some of the Internet improvement could be due to better performance by MIDS’s local ISP as well as increased port connection speeds by a number of distant sites. Nevertheless, Quarterman thinks that the 30 percent improvement is hard to explain solely by these fac- tors. But, does the data mean "all is cool with the Internet?" "No" says Quarterman. The "long latencies shown in the reports are real: servers that can’t keep up with traffic are a problem in themselves. And there really are fre- quent outages and breaks in wide area IP carriers and in their interconnecting points. But overall these don’t appear to be any worse than they used to be and the general trend is clearly towards improvement. To learn more, read John S. Quarterman, "Imminent Death of the Internet," Matrix News, June 1996 (http://www. raids, org). hardware and software, and compare the performance of one portion of the Internet to another. Major business users have also begun to lobby for lnternet performance statistics. Most big companies currently are unwilling to entrust strategic or mission critical business applications to the Internet. In some cases, they have per- suaded ISPs to establish dedicated IP intranets to handle such traffic so that guaranteed performance criteria can be met on an end-to-end basis. Some of these business users also want similar performance criteria for the public Internet so that they have wider service options. Traffic Measurements Why measure tnternet traffic? On the telephone network, the answer is obvious: traffic is money. Long distance (and some local) calls are billed by the minute. On the Internet, however, most users pay a fiat monthly fee; the cost is the same whether one is connected for a minute or an hour, and whether one downloads one thousand bytes or one billion. But, for Internet service providers, each user’s connect time and the number of bytes that are picked up and delivered does have economic implications. Computer hard drives, routers and private lines to backbone networks cost money. And the investment needed to maintain a given level of per- formance depends, in part, upon traffic volumes, although the relationship is much looser than on the public telephone network. To recoup the cost of these resources, ISPs will soon need to charge heavy users of the Internet--and those requiring real-time traffic delivery (e.g., for telephony}- more than users which place limited demands on the net- work. Traffic statistics are likely to play a role in making these new pricing mechanisms work. © TeleGeography, Inc. 1996 lntemet Traffic Statistics: An Overview Measuring Internet traffic volumes is not easy though. The Internet is a "connectionless" network. There is no fixed path for sending packets between one ISP and another, let alone between the ISPs in one country and another A sender’s message is digitally encoded, split up into pack- ets, and sent to the local ISP for onward transmission. The ISP then switches the packets to a backbone network by the most efficient route available using an ever changing set of routing tables. The backbone operator, in turn, forwards the packets in a like manner to another network and so on until they reach their destination. All point-to-point measurements of lnternet traffic volumes and traffic balances are therefore problematic. We may know how much traffic is switched from one network to another but the ad hoc muting of national as well as inter- national traffic may make ISP-by-ISP as well as country-by- country traffic statistics (i.e., in/out balances) quite unreli- able. Internet engineers also have found that the statistical tech- niques developed in the telephone industry for predicting the volume of traffic which can be handled by a given num- ber of dedicated circuits (e.g., Erlang-B statistics) and the distribution of calls in a given time period generally do not hold for the Internet’s connectionless traffic. For the most part, Internet traffic directed to a given address arrives, but sometimes it doesn’t or is delayed, and the arrival time and route cannot be predicted, on average, with any certainty. In sum, despite their bravado, it appears that most Internet operators are probably building blind. Backbone providers and local ISPs are adding massive new routers and large increments of transmission capacity. Yet, despite these large expenditures, no single tSP can be sure that any given investment will lead to a measurable increase in the quality F nternet Host Growth, 1981-1996 ¯ Internet hosts ~, Web servers 2,881,000 ¯ 00° 230,000 ¯ ¯ 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1998 100,000,000 10,000,000 1,000,000 100,000 10,000 1,000 100 In years past, each host on the Internet represented one computer. But the definition of hosts has changed such that a single computer can act like many hosts with many names and many addresses all at once. Network Wizards, the com- pany trying to keep track of Internet host growth, admits that "it is not possible to determine the exact size of the Internet." Source: Network Wizards, http://vwvw.nw.com/zone/host-count-history. Mathew Gray, a graduate student and researcher at MIT’s Med~a Lab, has been counting Web servers since the begin- ning. His most recent estimate puts Web growth on a pace that doubles the number of servers every six months. The Internet as a whole (measured by hosts), however, only appears to double every twelve months. Source: Mathew Gray of the Massachusetts Institute of Technology, http://www.mit.edu/people/mkgray. ©TeleGeography, Inc. 1996 TeleGeography1996/9-1 © TeleGeography, Inc. 1996 of service offered to users. Nor can the ISP be sure that its own efforts will reduce the likelihood of a future network wide "brownout" or crash. The three articles which follow look at different aspects of the Internet traffic dilemma described above. The first arti- cle by K. Claffy, a staff scientist at the San Diego Supercomputer Center, reviews the basic demand for Internet traffic statistics and other metrics; the technical problems involved in making the necessary measurements; and what might be done about these problems. The second article, by TeleGeography’s technical consultant, Zachary Schrag, looks at how new software deployed by ISPs might cope with the increasing volume of traffic generated by Internet telephony and other real time applications. A third article by Robert Shaw, an expert with the International Telecommunication Union (ITU) in Geneva, looks at the current debate over the assignment of address- es to Internet host computers. On first impression, this topic may not seem directly related to the size and measurement of Internet traffic volumes. But any regime for assigning Internet addresses affects both the number and location of networked computers. The location of computers in turn affects the volume of traffic between one part of the network and another. Hence, the ongoing debate over cyberspace property rights (i.e., registration of domain names) may well have a long term operational impact on future Internet traf- fic flows and the economic base of major ISPs. ¯ For further research: Much of the press about the coming Internet "crash" has been sparked by the views of Robert Metcalfe (http://www.c~o.com), ~nventor of Ethernet and a founder of 3¢om He is also a monthly columnist for InfoWorld. A good introduction to the claims and counterclaims on Internet congestion is prowded by the two-part series written by Charles Bruno ("lnternet Health Report: Condition Serious") for NetworkWorld beginning with the September 16, 1996 issue (http://www.nwfusion.com). Another overview appears in the October 1996 Network World issue headlined "The Big Crash--Is it Coming." See also "Bandwidth, Blockages, Brownouts," Meme 2.01 (http://www.virtualschool.edu/mon/lnternet) ; Jeremy Schlosberg, "It’s the Bandwidth, Stupid...or is it? Computerworld, April 29, 1996 (http://www.computerworld.com); Simson Garfinkte, "Web Brownout," Wired, September 1996, pp. 94-100. A sampling of operational statistics for the Internet is available from the Routing Arbiter (RA) project (http://www.ra.net) managed by Merit, a non-profit research and educational organization based in Michigan and funded by the U.S. National Science Foundation (NSF). However, some technical background ~s necessary to make sense of RA’s statistics, (e.g., "the RA ~s regularly performing ring tests across the NAP [Network Access Point] fabric and recording SNMP [Simple Network Mail Protocol] calls of the route servers for BGP [Border Gateway Protocol] statistics"). For a plain English guide, see Matrix Maps Quarterly (http://www.m~ds.org). The U.S. National Laboratory for Applied Network Research (NLANR) also maintains links to operational statistics data from research sites and ISPs (http://www.nlanr.net). Background on Internet pricing and congestion can be found at the following sites: http://www.sims.berkeley.edu/resources/infoecon/ pricing.html and http://www.oecd.org/dsti/gd_docs/s96_xxe.html. Also, recent work at the University of Texas at Austin on usage- based pricing has drawn attention (visit http://cism.bus.utexas.edu/ alok!pricing.html). © TeleGeography, Inc. 1996 Closing the Internet Statistics Gap CLOSING THE INTERNET STATISTICS GAP by K. Claffy This article covers three areas: ¯ why current Internet statistics are limited and why data collection is more difficult on the Internet than on the public telephone network; ¯ who needs Internet statistics and why and; ¯ possible models for ISPs and users to narrow the gap in understanding the nature of Internet traffic. How We Got Here: Limitations of Current Statistics The existence of the NSFNet (1986-1995) as a central net- work for the research and education community facilitated research into aspects of aggregate network traffic patterns and the anomalies in those patterns caused by the intro- duction of new or unique applications. Decommissioning the NSFNet backbone left the Internet community with no dependable public source of statistics on Internet work- loads. And yet the empirical investigations of current work- loads and their resource requirements, as well as how they change over time, is vital to supporting Internet evolution. Workload profiles are changing more rapidly than ever before. Keeping pace with them in an increasingly competi- tive, increasingly proprietary environment, is even more important now than during the life of the NSFNet backbone. The transition to the new NSFNet program, with commercial operators providing both regional service as well as cross- service provider network access points (NAPs--see page 51), renders statistics collection a much more difficult task. There are several dimensions of the problem, each with their own cost-benefit tradeoff. ¯ Thus far users have demanded few statistics from their providers. Even the NSF, one of the largest and most for- ward-looking users, called for few statistics in the coopera- tive agreements with NAP providers. This is understandable, because the NSF did not know enough about the way the post-NSFNet system would operate (neither did anyone else, although presumably the providers knew slightly more than the NSF). The situation is similar for other emerging lnternet service providers (ISPs). However, as it turned out, the NAPs and ISPs found it challenging enough just getting and keeping their infrastructure operational; statistics have never been a top priority. Nor do the NAPs really have a good sense of what to collect, as all of the technology involved is quite new to them as well. The issue is not whether traffic analysis would help, even with equipment and routing problems, but that traffic analysis is perceived as a secondary issue, and there is no real mechanism (or spare time) for collaborative development of an acceptable model. ¯ Many emerging Internet services are offered by com- panies whose primary business has been telecommunica- tions. The NAPs and the very high speed backbone network service (vBNS) providers are good examples. Phone com- panies are accustomed to reasonable analytic tools for mod- eling telephony workload and performance (e.g., Erlang B distributions).1 Unfortunately, the literature on Internet traffic characterization, both in the analytical and perfor- mance measurement domains, indicate that wide area net- working technology has advanced at a far faster rate than has the analytical and theoretical understanding of Internet traffic behavior.2 Moreover, there is still no consensus on how statistics can support research in IP traffic modeling. Critics within the lnternet community are skeptics of empirical studies that rely on collecting real data from the tnternet. These critics claim that because the environment is changing so quickly, within weeks any collected data is only of historical interest. They argue that research is better served by working on mathematical models rather than by empirical surveys that, at most, capture only one stage in network traffic evolution. ¯ The Internet’s early financial history has also con- tributed to the lack of traffic statistics. A few U.S. govern- ment agencies assumed the financial burden of building and maintaining the transit network infrastructure. There was little need to trace network usage for the purposes of cost allocation. The new commercial Internet is characterized by hundreds of ISPs, many on shoestring budgets in low margin competi- K. Claffy, Ph.D., is Assodate Staff Scientist at the National Laboratory For Applied Network Research (NLANR), San Diego Supercomputer Center, California. Correspondence should be directed to kc@sdsc.edu. Portions of this article draw upon "Report Of The NSF-Sponsored Workshop On Internet Statistics Measurement And Analysis, 19-20 February 1996"by Mark Garrett and Hans-Werner Braun et al., http://www.nlanr.net/ISMA/Report, and "A Survey of Internet Statistics/Metrics Activities" by T. Monk and K. Claffy, http://www.nlanrnet/metricsurvey.html. 31 TeleGeography 1996/97 © TeleGeography, Inc. 1996 tion. They generally view statistics collection as a luxury that has never proven its operational utility. The last publicly available source of lntemet workload and performance data for the NSFNet backbone, was basically a gift from the NSF--an investment of U.S. tax dollars with the hope that tools, methodologies, theories of traffic, refinements and feedback would emerge from groups like the Internet Engineering Task Force (see Box 2). But there was never any fiscal pressure to justify allocating the resources required to collect statistics. Larger telecommunications companies entering the market- place will inevitably learn to devote more attention to this area. But the pressure will probably not occur until the system breaks, at which point billed customers will demand, and be willing to pay for, better guarantees and data integrity.3 ¯ Privacy has always been a serious issue in network traffic analysis. Most ISPs have service agreements that pro- hibit them from revealing information about individual customer traffic. Collecting and using more than aggre- gate traffic counts will require customer cooperation regarding what may be col- lected and how it will be used. For an ISP to breach customer expectations or ethical standards, even for the most noble research goals, does not bode well for future business. mance with other networks or with a baseline. Increasingly, both users and providers need information on end-to-end performance, which is beyond the realm of what is control- lable by individual networks.5 The vacuum created in national-level statistics/metrics col- lection after the transition to the commercial architecture has also complicated planning for national service providers and others. Detailed traffic and performance measurements are essential to identifying the causes of network problems and formulating corrective actions. But it is trend analysis and accurate network!systems monitoring that permit net- work managers to identify hot spots (overloaded paths), THE NEW INTERNET IS CHARACTERIZED BY HUNDREDS OF ISPs, MANY ON SHOE- STRING BUDGETS... THEY GENERALLY VIEW STATISTICS COLLECTION AS A LUXURY THAT HAS NEVER PROVEN ITS UTILITY. OPERATIONAL ~J Where We Are: Who Needs Internet Statistics and Why In Routing in a Multi-Provider Internet, Yakov Rekhter, an engineer at Cisco Systems, writes: Despite all the diversity among the providers, Internet-wide IP connectivity is realized via Internet-wide distributed routing, which involves multiple providers, and thus implies a certain degree of cooperation and coordination. Therefore, we need to balance the provider goals and objectives against the public interest of Internet-wide connectivity and subscriber choic- es. Further work is needed to understand how to reach the balance. 4 predict problems before they occur, and avoid them through efficient resource deployment and optima! network con- figuration.6 Service Quality and Pricing Models Demands for implementing multiple Internet service levels are increasing. From the providers’ standpoint, such offerings will enable increased revenue through business-quality services offer- ings. From the user’s perspective, the ability to contract for higher priority ser- vice will enable many industries to switch from intranets and private net- works to Internet-based infrastructure. The ability to specify or reserve the ser- vices one needs from the network will in turn require mechanisms for accounting and pricing (or else there is no incentive not to reserve all one can, or not to use the highest priority).7 The Internet is still relatively devoid of pricing models or other mechanisms to allocate and prioritize scarce resources--particularly bandwidth--and acutely needs mechanisms for more rational cost recovery, that is, more accurate accountability for resources consumed. In particu- lar, the Internet architecture is not prepared to deal with traffic flows that vary by several orders of magnitude. Of major concern in workload profiles today is the disparity in size between most current lnternet flows/transactions, at less than ten packets, and newer multimedia applications with much higher volume and duration. Many internet service providers currently collect basic sta- tistics on performance of their own infrastructure, typically measuring utilization, availability, and perhaps delay and throughput. In the the post-NSFnet era, the only baseline against which networks evaluate performance is their past performance metrics. There are no data available or even standard format defined against which to compare perfor- The disparity in workload profiles in the current cross-sec- tion of lnternet applications necessitates revised metrics of network behavior. Simple mean or peak utilization figures do not address a service provider’s engineering needs, because they say nothing about the transaction profile constituting and perhaps dominating those figures. Keeping track of workload profiles requires measuring flow data at relevant network locations.8 © TeleGeography, Inc. 1996 Closing the Intemet Statistics Gap Box 1. The NSF Transition Until recently, the vast majority of Internet hosts were/n the United States and relied upon a common back- bone network supported by the U.S. National Science Foundation (NSF). The NSFNet was the largest and most widely used Internet interconnection facility and backbone traffic measurements provided a reasonable indication of Internet traffic trends worldwide. The NSF decommissioned this government funded backbone in April I995, as it became clear that multiple commercial providers were in a position to offer Internet backbone services. NSF withdrawal from support of backbone services to the research and engineering community involved modifications to the NSFNet architec- ture to ensure Internet stability dunng the transition from government supported services to full privatization of the network. These modihcations involved the creation of four new projects, three/nfrastructural, and one research related: ¯ general purpose Network Access Points (NAPs): to connect the commercial backbone networks thus avoiding network partitioning; ¯ a routing arbiter, to provide routing coordination among providers during the transition fsee http://www.ra.net) ; regional/nterconnectivity to regional networks. (Spec/hcally, NSF-sponsored regional providers, i.e., those who received funding from the NSF throughout the life of the NSFNet, will continue to receive funding for four more years so long as they connect to a backbone provider that connects to all three NSF NAPs. This constraint is the only leverage NSF had to prevent partitioning since the backbone providers themselves received no funding from NSF and thus had less incentive to "do the right thing" at the time. In any event, regional funding ends after four years, at which point the regional providers will have had ample opportunity to become fully self-sustaining w/thin the marketplace.) ¯ very High Speed Backbone Network Services (vBNS), a wide area network initially connecting the NSF supercomputer centers for use by both application scientists as well as network researchers. More accurate resource consumption statistics, and con- comitant pricing models, will allow progress with another severe need in the current infrastructure: a service architec- ture from the perspective of the end user. Maximizing value for the end user in the Internet is difficult since the economic value model is quite randomized. In most markets, prices rise along with quality of service (QoS). The Intemet should be no exception: rational pricing would provide the right feedback to providers and users to encourage more appro- priate use. Currently QoS signals are unclear. Users need operating signals, accompanied by measurement-distin- guishable service qualities, so they can declare the QoS for which they will pay. Otherwise high value users may get degraded by high requirement, low value users.9 The lack of a common business model for inter-ISP relations inhibits settlements. Some suggest that the per minute traf- flc settlements used by international telephone carriers could be applied to ISP settlements. But because the Internet relies upon a connectionless network, where per minute point-to-point traffic flows are meaningless, others argue that new pricing models must be developed. Where to Go Next: IWodels for Data Collection Although much of the Internet community agrees that we should seek out measurement statistics in the commercially decentralized lnternet, there is definite dissonance as to which measurements would help, and who should have access to them. While a public measurement infrastructure could help researchers and end users, ISPs might benefit more from the ability to collect statistics that were too sen- sitive to release publicly, and perhaps from comparing them to corresponding statistics from other ISPs. The best means for collecting the various statistical needs outlined in this article may be a provider consortium. The National Laboratory for Advanced Network Research (NLANR] has suggested a possible framework for such a col- laboration, which would serve as a forum for: TeleGeography 1996/97 © TeleGeography, Inc. 1996 ¯ facilitating the identification, development and deployment of measurement tools across the lnternet; ¯ providing commercial providers with a neutral, confidential vehicle for data sharing and analysis; ¯ providing networking researchers and the gener- al Internet community with additional realtime data on Internet traffic flow patterns; ¯ enhancing communications among commercial Internet service providers, exchange/peering point providers and the broader Internet community. Focal areas of the consortium would include tracking out- ages, congestion monitoring, examining the use of backbone routing, studying peering relationships, and creating a forum for the discussion of charge policies. Market pressures upon ISPs to participate in such a consortium concept include: ¯ customers’ increasing dependence on the Internet for mission critical applications; ¯ settlements that require authenticated and pos- sibly confidential provider statistics; ¯ the meshed nature of the Internet, which sug- gests that no single company can do it alone, and that systemic improvements will require collabora- tion. The business constraints hindering such cooperation relate to the competitive nature of the Internet business environ- ment, as well as the appearance of industry collusion by major providers. However, a charter with principles of open- ness and inclusion can readily address these concerns, as well as address constraints arising from the lack of adequate pricing models and other mechanisms for economic ratio- nality in Internet business practices. Technology constraints hindering the collection and analysis of data on Internet metrics center on the nascent develop- Table 1. Internet Metrics and Tools Needs Why Measurements and Tools Where Aggregate traffic flow analysis Capacity/topology planning Traffic matrices At a sample of ISPs Trouble-shooting ¯ Cache management ¯ Full header traces At a few high aggregation ¯ Optimize queuing points (e.g., near NAPs) ¯ Congestion and scahng dynamics ¯ Aggregate transport behavior Specifications for ISPs to g~ve vendors of routers Router benchmarking ¯ Flow counts/parameters ¯ NAPs ¯ Backbone core points ¯ corporate campuses ¯ Large customers Specifications for users to give ISPs ¯ Serwce quality assessments ¯ Comparison of ISPs ¯ Billing ¯ Transit-related settlements agreements Multi-function beacon testing platforms ¯ Mathis’ treno ¯ Paxson’s probe daemon ¯ Bridged at strategic interconnects ¯ Customer sites Routing stabihty (i.e., avoiding crashes; "brownouts") ¯ Reliabihty ¯ Availability ¯ Quality of service ¯ Outage tracking ¯ Route archiving ¯ Caching ¯ Database ~mplementat~on Route servers Testing new apphcat~ons/protocols ¯ Deployment of IPv6 ¯ Bandwidth reservation ¯ Multicast caching ¯ Directory services ¯ Interoperability tracking across ISPs ¯ Internal and border routers Peering points © TeleGeography, Inc. 1996 Closing the Internet Statistics Gap Box 2. The Internet Engineering Task Force The Internet Engineering Task Fame (IETF) is a large, open community of network designers, operators, ven- dors, and researchers whose purpose is to coordinate the operation, management and evolution of the Internet, and to resolve short-range and mid-range protocol and architectural issues. It is a major source of proposa/s for protoco/ standards which are submitted to the Internet Architecture Board (lAB) for fina/ approval. The IETF meets three times a year and extensive rn/nutes are included in the IETF Proceedings, The IETF IP Provider Metrics (IPPM) working group is one of many working groups in the IETF, it is comprised of researchers and service providers interested in defining basic metrics and measurement methodologies in order to develop standardized performance evaluations across different Internet components. merit stage of measurement tools for IP and ATM (Asynchronous Transfer Mode) network flows. Complications also arise from the adoption of new and emerging technolo- gies (e.g., gigaswitches and ATM). These and other con- straints are likely to be solved, given sufficient technical attention and market pressure.Developing an effective provider consortium would require, minimally: ¯ participation by three or more of the major ser- vice providers (e.g., ANS, AT&T, BBN Planet, MCI, Netcom, PSI, Sprint, or UUNet); ¯ a neutral third party with sufficient technical skills to provide the core data collection and analysis capabilities required by the consortium; ¯ appropriate privacy agreements to protect the interests of members; ¯ agreement on which basic metrics to collect, col- late/analyze, and present and; ¯ agreement on which tools to develop, particular- ly those related to emerging infrastructures using new technologies. This organization would coordinate not only a solid consis- tent library of tools that could appeal to both users and providers, but, with the cooperation of ISPs, would allow the testing of real data without compromising anyone’s propri- etary data or technology. Data collection would strictly focus on engineering and evolution of the overall Internet envi- ronment. Accurate data on traffic patterns could allow engi- neers to design more efficient architectures, and design them more quickly, thus conserving both labor and resources now unnecessarily allocated to parts of the net- work where they are not needed. The right statistics collec- tion and cross-ISP dissemination mechanisms would facili- tate faster problem resolution, saving the time and money now devoted to chasing down bugs. Developing the appropriate metrics and tools to measure such phenomena, as well as end-to-end performance and workflow characteristics, is still a Looming task. Experience with data will foster the development of more effective usage-based economic models, which, in the final analysis, will allow ISPs to upgrade their infrastructure in accordance with customer demand. ¯ TeleGeography 1996/97 © TeleGeography, Inc. 1996 Endnotes 1. Erlang-B tables are used by telephone engineers to determine the number of circuits needed to accommodate a given volume of traffic at a determined grade of service (i.e., to determine whether a call will be blocked). The tables reflect the work of A. K. Erlang, a Danish mathematician. Eriang correctly assumed that for most telephone networks the number of calls that terminate in any given time period is proportional to the elapsed time and to the number of calls in progress but does not depend on when the calls were ini- tiated. He assumed a Poisson distribution of calls arriving in a given interval A Poisson distribution is bell shaped but the peak is shifted offmcenter (e.g., a Poisson distribution would show that the probability a given number of calls will be dialed in any 12 second period is h~ghest between four and five seconds.) See generally I. R. Pierce, Signals: The Telephone and Beyond (W.H~ Freeman & Co., San Francisco, 1981) pp. 133-134. 2. Traditional mathematical modeling techniques, e.g., queuing theory, have met with littte success in today’s lnternet environ- ments. Years ago, for example, the assumption of Poisson arrivals was acceptable for the purposes of characterizing small LANs. On the lnternet, however, whether in terms of packet arrivals within a connection, connection arrivals within an aggregat- ed stream of traffic, or packet arrivals across multiple connections, collected data do not fit a Poisson distribution. Some experts are investigating alternatives to Poisson modeling, specifically the use of self similarity (fractal) mathematics to model IP traffic. See V. Paxson, "Empirically-Derived Analytical Models of Wide Area TCP Connections," IEEE/ACM Transactions on Networking, Vol. 2 No. 4, August 1994; also see W. Leeland, M. Taquu, W. Willinger, and D. Wilson, On the Self-Similar Nature of Ethernet Traffic (extended version)," in IEEE/ACM Transactions on Networking, Vol. 2, February 1994. For prior stud- ies on national backbone traffic characteristics see the following sources: K. Claffy, H.mW. Braun, and G. C. Polyzos, "Long-term traffic aspects of the NSFNET," in Proc. oflNET "93, pp. CBA--I:I 0, August 1993; K. Claffy, H.~W. Braun, and G. C. Polyzos, Tracking long- term growth of the NSFNET backbone," Proceedings of the ACM, vol. 37, pp. 39-52, August 1994; K. Claffy, Internet workload char- acterization, Ph.D. thesis, UC San Diego, June 1994. The limitation of this work has led to a review of the WAN traffic characterization which focus on a single or a few attachment points to transit networks to investigate shorter-term aspects of certain kinds of Intemet traffic (e.g, TCP, UDP, and DNS). 3. With the transition to Asynchronous Transfer Mode (ATM) and high speed switches, it is in many cases no longer even be technically feasible to access IP layer data in order to do traffic flow profiling, certainly not within commercial ATM network equipment. Many switches have little if any capability for collecting statistics at NAPs, or even looking at traffic in the manner allowed on a broadcast medium (e.g., FDDI and Ethernet), where a dedicated machine can collect statistics without interfering with packet forwarding. Statistics collection functionality in newer switches takes resources directly away from forwarding, dri- wng customers toward switches from competing vendors who sacrifice such functionality in exchange for speed. 4. Yakov Rekhter, "Routing in a multi~provlder Internet," Internet Request for Comments Series RFC 1787, April 1995. 5. Another example of statistics maintained in isolation within an individual ISP is trouble ticket tracking of problems that origi- nate and are resolved within the context of a single ISP. Throughout most of the life of the NSFNet backbone, resolving route instabilities and other trouble tickets was the the responsibility of Merit, which had a cooperative agreement with NSF for oper- ation of the NSFNet backbone. In the current environment there is no such entity to claim to share responsibility for national much less global management of the lnteroet. As a result, there are no scalable mechanisms available for resolving or tracking problems originating or extending beyond the control of an indiwdual network. Route instability is another area that can have a direct, sometimes profound, effect upon the performance of individual net- works. Some networks are seeking to improve the stability of their routing by peering directly with the Routing Arbiter (RA) at network access points (e.g., SprintNAP and FIX-West/MAE-West) (see Box 1 to the main article). 6. Examples include measurements of: round-trip-time (RTT), e.g., with probe queries, to assess congestion and other condi- tions at an infrastructure-wide level; routing behavior (beyond that currently available through the Routing Arbiter project), to assess status and stability, as welt as unusually configured routing and the conflict between simultaneous presence of more spe- cific routes for a given route aggregate. 7. Some fear pricing will stifle the open, vibrant nature of the lnternet community. But pricing is likely to motivate the con- structive exploration of more efficient and innovative networking applications. 8. NLANR currently supports http://www.nlanr.net, an interface to the operational collection of such data at several points, including data from the FIX-West multiagency network interconnection facility. 9. In his recent "lnternet Draft on Metrics for lnternet Settlements," Brian Carpenter (of CERN) asserts that financial settle- ments are a "critical mechanism for exerting pressure on providers to strengthen their infrastructures." He suggests that met- rics used in Internet settlements should not rely on expensive instrumentation such as detailed flow analysis, but rather simple measurements, estimated, if necessary, by statistical sampling. © TeleGeographyo Inc. 1996 The Achilles Heel of Internet Telephony THE ACHILLES HEEL OF INTERNET TELEPHONY New Congestion Controls Could Raise the Price of Realtime Traffic by Zachary M. Schrag nternet telephony--using a computer to make telephone calls over the lnternet, a private network--is cheaper than public telephony for three reasons. First, it uses net- work resources more efficiently, by relying on a packet- switching technology rather than circuit-switching, as does the public network.1 Second, it generally avoids the economic regulations which constrain the public network. In the U.S., for example, Internet telephony escapes payments to universal service funds and other local interconnection fees. Finally, it exploits the current architecture and billing practices of the Internet, using more resources than it pays for. Today, with the right software, a microphone and speakers, and an Internet connection, one can talk with a similarly equipped friend or colleague an ocean away without paying long distance charges per-minute or per-packet.2 For the most part, users of Internet telephony are not charged more than other Internet users, though they may consume very much more capacity and contribute disproportionately to network congestion. But this may soon change. Before unmetered lnternet tele- phony becomes attractive to millions of users, it may be sharply curtailed by Internet service providers (ISPsl anx- ious to conserve their expensive capacity. New software pro- tocols now coming into use will give the ISPs the ability to limit or snuff out Internet calls on their networks with just a few keystrokes. Once these protocols are in place, Internet telephony and other realtime applications will run only at the mercy of the ISPs. To understand why, one must look at how these appli- cations differ from e-mail, World Wide Web traffic, and other Internet applications. Not all Internet packets are alike, and the same distinctive protocols that Internet tele- phony uses to push its packets ahead of the crowd may also allow ISPs to filter them out. TCP vs. UDP In one sense, protocols are the Internet. In 1974, Vinton Cerf and Robert Kahn created the lnternet as we know it by designing TCP, or the Transport Control Protocol. TCP was designed to carry the most popular forms of data at that time: e-mail messages and files to be transferred. But it has since been adopted for Usenet, gopher, and the World Wide Web, I the last of which accounts for about two-thirds of all traffic on the lntemet (see Figure Cerf and Kahn decided that the most efficient way of trans- mitting these types of data would be to break them into packets and treat them as store-and-forward messages. Their speed in getting across the network was less important than the assurance that they would arrive eventually. To ensure that its packets would get through even when the lntemet was clogged, TCP was later modified to include Slow-Start. When faced with Internet congestion, Slow-Start TCP packets wait their turn, traverse the network by differ- ent routes if necessary, and then are reassembled at their destination. If packets are lost, they are sent again until they eventually arrive. Thus, if necessary, TCP sacrifices speed for reliability. In 1979, Jonathan Postel created an alternative to TCP, called the User Datagram Protocol.3 Because UDP sends only single, small packets at a time, it skips the reassembly process required by TCP, making it simpler and faster than TCP, but less reliable. If some UDP packets cannot get to their destination in the specified time, they are never deliv- ered. Unlike TCP, UDP does not delay its packets when confronted by congestion on the network. As a result, today’s routers favor UDP packets over TCP packets when congestion hits. They permit UDP packets to jump to the front of the line, not because anything in the UDP specification said they should, but simply because UDP was commonly used for sending important, brief messages like Internet Name Server queries. Though UDP was not designed as a high-priority protocol, in practice it functioned that way until very recent- ly. Zachary M. Schrag is a technical consul- tant to TeleGeography and a Ph.D. candi- date at Columbia University. He can be reached at zms2@columbia.edu or http://www, columbia, edu/-zms2. Realtime UDP In 1992, the Internet Engineering Task Force’s Audio/Video Transport Working Group was chartered to explore ways to transmit realtime data over the lnternet (primarily for con- ferencing, rather than one-to-one conversations). The group realized that because of UDP’s impatience, it might be used to carry time-sensitive data like tele- phone calls. But the group doubted that UDP would be the best tool for the job, saying in its charter that "UDP transmission of audio and TeleGeography 1996/97 © TeleGeography, Inc. 1996 video is only sufficient for small-scale experiments over fast portions of the Internet.’’4 As the lnternet grew, the T-1 (1.5 Mbps) lines that had seemed fast became common, and using UDP for real-time transmission became more practical. Hence, when the work- ing group released its specifications for the Realtime Transport Protocol (RTP) in January 1996, it stated that applications would "typically run RTP on top of UDP" (that is, with UDP) although "RTP may be used with other suit- able underlying network or transport protocols. ,,5 Meanwhile, realtime applications using proprietary proto- cols on top of the UDP layer began to enter the market- place. These applications included popular lnternet telepho- ny packages, like lnternet Phone, as well as CU-SeeMe, a videoconferencing package developed at Cornell University and now commercialized by White Pine Software. Each of these applications depend upon UDP’s speed, compensat- ing for UDP’s tendency to drop packets by interpolating lost phonemes or video frames. The more realtime applications were sent over the lnternet using UDP, the more the software’s main drawback came to the fore: UDP lacks congestion control. It provides no incen- rive to end-users to conserve bandwidth. Thus as RTP’s designers put it, in January 1996: [T]he current Internet cannot yet support the full potential demand for real-time services. High-band- width services using RTP, such as video, can poten- tially seriously degrade the quality of service of other network services. Thus, implementors should take appropriate precautions to limit accidental bandwidth usage. Application documentation should clearly out- line the limitations and possible operational impact of high-bandwidth real-time services on the Internet and other network services. 6 So far, realtime UDP software has caused congestion only on relatively small sections of the Internet, or for short lengths of time. But the effects of such "brownouts" fall not on the UDP users who cause them, but on TCP-using bystanders on the same network. According to Jon Knight of Loughborough University of Technology, some ISPs are reticent about high bandwidth appfica- tions. Firstly, these applications typically use UDP based technology (and in the case of many of the pro- prietary Mac and PC based ones, unicast UDP). High UDP traffic flows can be bad news for other users as, unlike TCP, UDP doesn’t have any way to ’back off" orld Wide Web Traffic as a Percentage of Total Internet Packets 8O Figure 1: HTTP (World Wide Web) Packets as a Proportion of Total Internet Packets on the ANS backbone. For now, the Web is by far the most popular use of the Internet, responsible for much of the attention, money, and subscribership the Net has received. ISPs would be foolish to let a few Internet telephony users degrade the service that most of their subscribers use. Source: Daniel McRobb, ANS Network Services ©TeleGeography, Inc. 1996 © TeleGeography, Inc. 1996 The Achilles Heel of Internet Telephony when it hits congestion in the network. What happens in some situations is that a couple of users start high bandwidth UDP sessions which cause the better behaved TCP connections opened by lots of other users to back oK The other users then see rotten per- formance and complain to the ISP. 7 As one group of researchers put it, current "congestion con- trol designs do not work if some users choose to misbehave; in particular, greedy users can capture more than their share of bandwidth by not responding to congestion signals. Such greedy users not only capture more bandwidth for them- selves, but also seriously degrade the service obtained by cooperating users. ,,8 Congestion Control UDP packets only precede TCP packets because routers, the computers that pass on Internet traffic, let them. And routers can be reprogrammed. To reduce congestion from voice or video, routers need only do two things: identi- fy voice and video streams and act to limit these streams. Identification is quite straightforward. Despite occasional claims that lnternet telephony packets are indistinguishable from other Internet traffic, in fact each lnternet telephony software package marks its packets with a distinctive port number. For example, packets generated by Internet Phone all are marked with port number 22555. Such crude, all-or-nothing means of congestion control are necessary for the moment. The development of realtime software has simply outpaced the development of conges- tion control protocols. The current system for handling Internet traffic was designed for the small, cooperative, non- profit Internet of the 1970s and 1980s. That Internet still exists, but it has been submerged beneath the large, anony- mous, commercial Internet. DROPPING UDP PACKETS WOULD FREE UP BANDWIDTH FOR WEB SURFERS AND OTHER TCP USERS, AND IT WOULD DEGRADE THE QUALITY OF INTERNET TELEPHONY CONVER- DISCOURAGING SATIONS, THUS J MORE CALLS. ISPs beset by congestion (and customer complaints), can easily program their routers to reject packets based on their port numbers, in effect turning off specific applications. This has already happened. To prevent a small number of video users from taking over their expensive transoceanic capaci- ty, some providers in the Netherlands and some universities in Australia have already blocked UDP port use by CU- SeeMe, at least during certain hours. (Forty people receiv- ing CU-SeeMe streams can tie up an entire T-1 line, which could be an ISP’s entire link to the main Internet back- bones.) Thus, by denying UDP packets to a few, they keep the TCP packets flowing smoothly to the majority of their users. Because all the most crucial UDP applications (like the Name Server) have port numbers below 1024, an ISP des- perate to shake off realtime traffic can block all ports above 1024, disabling all realtime voice and video while preserv- ing the basic UDP functions. But ISPs will soon gain more control over UDP traffic. A new protocol, called Random Early Detection (RED), allows operators to ration bandwidth more carefully than before. RED, which ISPs can install on their touters, identifies offending packets not by reading port numbers but simply by determining which streams are con- tributing the most to congestion. Then it can clamp down on just those streams. "RED gateways could easily identify which connections have received a sig- nificant fraction of the recently-marked packets .... This information could be used by higher policy layers to restrict the bandwidth of those connections dur- ing congestion.’’9 In other words, RED could let by swift, slim messages like Name Server queries and http requests for files, while clamping down hard on bandwidth guzzlers like voice and video. Though RED was first proposed some years ago, only now is it coming into widespread use. According to Stephen Casner, chairman of the IETF working group on Audio-Video Transport, with the significant increase in UDP traffic ... the time has come when touters must implement congestion control. Algorithms such as RED... have already been implemented in routers and are ready to deploy. It is expected that network service providers will deploy the new software and turn on these congestion control mechanisms over the next few months. This will cause the loss experienced by UDP traffic to go wa~t up, and the loss experienced by TCP traffic to go down, so the TCP traffic will regain its fair share, lO Two leading manufacturers of routers have announced their support of the Integrated Services Architecture (ISA), which includes RED. 11 With RED, ISPs will be able to fine tune the use of their net- works. If a network manager decides that lnternet telepho- ny users are clogging the pipe, she can configure RED to begin dropping UDP packets. Dropping UDP packets would free up bandwidth for web surfers and other TCP users, and TeleGeography 1996/97 © TeleGeography, Inc. 1996 it would degrade the quality of Intemet telephony conver- sations, thus discouraging more calls, Pricing the Internet Of course, ISPs do not want to punish their customers; they just want to make sure that each user pays his share. They will be able to use RED to force users into a new pricing regime, in which delay-tolerant and realtime users pay for the level of service they want. The key to such a regime is yet another protocol, the Reservation Protocol (RSVP). RSVP allows providers to charge additional fees for guaranteed bandwidth. Only those realtime UDP streams paying the extra toll would be allowed through the RED gateway.12 This would mean that lnternet telephony would no longer be unmetered. It might remain quite cheap; anywhere from a penny per minute to US$0.50, depending on a host of vari- ables, but not free. And cheap and free can be very differ- ent when it comes to business models. Given the range of international telecom services available, it is not clear why consumers would pay much for Internet telephony when other inexpensive options, like call-back, offer superior quality and convenience. Alternatively, the current all-or-nothing approach to block- ing UDP could divide providers into two classes. Some providers could raise their fixed monthly fees, using the money to invest in more capacity for realtime applications, thus forcing realtime users to bear the full cost of their bandwidth demands. Other providers could block out the realtime UDP traffic, keeping congestion down without rais- ing prices. Whether they choose between providers or between two types of service from the same provider, users will be faced with two classes of Internet services: a more expensive real- time service and a less expensive delay-tolerant service. In a February 1996 interview with Interactive Age, Vice President Marc Andreessen of Netscape (a company pro- moting lnternet telephony) himself predicted that the spread of Internet telephony would destroy the very struc- ture of unlimited-use pricing that has helped make it so alluring. Andreessen predicts extra fees for those using real- time lnternet services, such as voice and video. ’~kt the end of the day, it is an economic question. Either it will be worth the additional money (for end users) or it won’t. Prices will be adjusted to reflect that reality." 13 Endnotes 1. Unlike circuit-switched telephony, which consumes an entire circuit even when people on both ends are silent, Internet telephony only uses bandwidth when someone says a word. Numerous technical articles on Internet telephony may be found at http://www.von.org, the Web site for Voice on the Net (VON). 2. Where local access to the Intemet is metered, however, per minute charges generally apply as they do for any on- line session. 3. Internet Engineering Note (IEN) 88, http://www.uwaterloo.ca:80/uw_infoserv/ien.html 4. AVT Charter, http://www.ietf.cnri.reston.va.us/html.charters/avt-charter.html 5. RFC 1889, ftp://ds.internic.net/rfc/rfc1889.txt 6. Id. 7. Personal communication, February 14, 1996 8. Christopher Lefelhocz et al., "Congestion Control for Best-Effort Service: Why We Need a New Paradigm." IEEE Network, January/February 1996, vol. 10, No. 1, http://www.ieee.org/comsoc/lefelhocz.htmt 9. Sally Floyd and Van Jacobson, "Random Early Detection Gateways for Congestion Avoidance," IEEE/ACM Transactions on Networking, August 1993. 10. Personal communication, July 2, 1996. 11. Fred Baker, "Real-Time Services for Router Nets," Data Communications on the Web, May 21, 1996, ht-tp://www.data.com/Tutorials/Real_Time_Services.html 12. See Thomas Nolle, "Reservations about RSVP," NetworkWorld, October 28, 1996. 13. Richard Karpinski, "Andreessen Discusses Netscape’s Real-Time Audio, Video" Interactive Age, Februaryll7,1996. © TeleGeography, Inc. 1996 Internet Domain Names INTERNET DOMAIN NAMES Whose Domain Is It Anyway? by Robert Shaw The past year has seen an intense debate in the press and on the lnternet regarding policies for allocating domain names, the Internet’s online equivalent of telephone num- bers. The outcome is likely significantly to affect how the lnternet is governed as well as its workload--that is, the routing of traffic from one section of the lnternet to anoth- er. The millions of computers which are connected to the lnternet all have a unique geographical location. But on the Internet, what matters is their domain name or virtual address. And today the most coveted place on the Internet is the .corn domain, a geographical domain intended for commercial organizations. It is now used for everything from locating the World Wide Web site of a company (www.ibm.com) to the online marketing of Hollywood movies (www.missionimpossible.com). Until recently, registration of such domain names was first come, first served. Many names corresponding with well- known trademarks have been registered by nimble ama- teurs, sometimes with innocent intent, and sometimes in the hope of a quick payoff from a sleepy company which suddenly wants to get online. But the corporate world (especially in the litigious U.S.) is waking up (see Box 2 below). Even so, registration under the .corn domain won’t work for much longer. This domain now has over 550,000 entries and registrations are running at over 20,000 per week; the desirable permutations of easy to remember names is quick- ly running out. With the current system, there’s not enough space for the millions of new registrants who will want their own mnemonic cybersite. Worse still, what is essentially a U.S. mess is turning into a global one. Domain name registrations in other countries are, unlike in the U.S., typically tagged with a "country code." For example, France uses .fr and Japan uses .jp. But .com is a non-country specific international top level domain that can be used by anyone in the world. As of June 1996, non-U.S, entities accounted for about 75,000 of the exist- ing international top level domain registrations--and their relative percentage is rising.] What will happen when more non-U.S, com- panies want in and start waving their national trademarks too? So the rules for the Internet Domain Name System (DNS) are clearly overdue for reform. However, there is little agree- ment on what to do next. The only consensus seems to be that something needs to be done--and fast. But what and how? Given the many policy issues relating to any modification of the DNS, this problem particularly begs the question: who should be setting Internet policy? Should it be the brilliant engineers who keep the Net running? Should it be govern- ments who have partially subsidized domain name registra- tion services? Should it be the commercial sector, which depends increasingly on the Net and is concerned about the relationship between trademarks and domain names? What about Internet service providers (ISPs) who also have a clear interest in a stable infrastructure? Should internation- al multilateral organizations try to provide a forum for a neutral and international solution? (See Box I at page 43, "The Internet’s Unelected Governors," for profiles of some of the major players.) And what about the money involved? lnternet domain names are a new form of intellectual property and insofar as they represent a scarce resource, they are quite valuable, both to those who hand them out and those who own them. Who, if anyone, should be profiting from the sale of this new cyber property? What is the Domain Name System? The Internet Domain Name System basically provides a method to map a user friendly name such as "www.microsoft.com" to a numeric address such as "198.105.232.6." The suggestion for a hierarchical name space for the Internet was probably publicly described first in the RFC2 document lnternet Name Domains by D.L. Mills of Comsat Laboratories in 1981.3 A few years later, J. Postel and J. Reynolds described a set of generic top level domains (TLDs) that are still in use today (see Table 1).4 Robert Shaw is an Advisor to the Information Services Department, International Telecommunication Union (ITU), Place des Nations, Geneva, Switzerland. Contact shaw@itu.ch. Postel and Reynolds also defined top level domains based on two-letter "country codes" from the International Organization for Standardization’s (ISO) 3166 stan- dards.5 Delegation and manage- merit of ISO 3166-based TLDs have been typically assigned to national, or (when not practical) regional reg- TeleGeography 996/97 © TeleGeography, Inc. 1996 Table 1. Generic Top Level Domains Generic Top Level Domains Defined in RFC 920 Intended Usage .gov government .edu education ,corn commercial .mil military .org organization Top Level Domains Later Defined Intended Usage .net networking providers .int international treaty organizations and Internet databases ISO Country Code Intended Usage Top Level Domains (According to Regional or (a few examples) National Registry Policy) .af Afghanistan .ch Switzerland .fr France .us United States of America .zw Zimbabwe Note: There is one more top level domain-- arpa--wh~ch ~s solely used for Internet techmcal infrastructure needs. istries.6 At the national level, these assignments have typi- cally been made on a first come, first served basis. The usage of ISO-3166-based TLDs introduces a geograph- ic and territorial (and therefore political) component to domain name space. However, since the lnternet grew in a bottom-up fashion from research and academic environ- ments, governments have traditionally taken little interest in the national registries. This is slowly changing. In some cases, government authorities have begun to give specific guidance to national name registries.7 Other governments (often those with a restrictive political or highly regulated communications environment) have insisted on total responsibility for TLD management. Recognizing the politics involved, it has been one of those many unwritten Internet policies that if a government requests TLD management for a country, it gets it.8 With the commercialization of the lnternet, national TLD registries have been rapidly moving away from the previ- ously informal arrangements for domain name registrations. For example, both the U.K. and the Netherlands9 have recently set up new legally distinct entities to handle domain name registrations. Naturally, the quickly developing rela- tionship between domain names and trademarks has also led registries to be more concerned about potential legal lia- bility. For example, the French Network Information Center (NIC) advisory group on naming policy recently declared: The majority of representatives judge that it is neces- sary to apply legal criteria, exterior to the France NIC, for the definition of the naming plan ... in a manner to avoid direct litigation. ]o Another trend of the national registries is the widespread introduction of annual fees for domain name holders. For example, Switzerland’s registration authority SWITCH intro- duced an annual fee of approximately US$80 after consul- tation with the Swiss telecommunications regulatory author- ity.11 In Australia, France, Japan, the United Kingdom and other countries, the national TLD authorities have created subdomains under the country code TLD to further classify and divide the name space.12 Although there is also a country-code-based .us domain,] 3 it is not used very much and many don’t even know it exists. Many complain there is a problem of access. The .us domain has an elaborate subdomain naming scheme based on "political geography," but noticeable by their absence are the .com, .net and .org subdomains that many other coun- tries use. In any case, the poorly-developed use of the .us domain and what Intemet cognoscenti refer to as "domain name envy," ".corn envy" or "international envy" has led individ- uals and corporations alike to register in the generic TLDs without a national tag. Some perceive this as a great "equalizer" which reflects part of the spirit of the Net. However, outside the United States, American’s lack of use of the .us country tag is widely perceived as an abuse of the domain name system. There is a wide spread misperception that the generic TLDs .com, .net, and .org are U.S.-only domains. In fact, the rest of the world is learning that anyone can register in these domains and an increasing number of non-U.S, companies are doing so--from Swissair (swissair.com) to Mercedes- Benz (mercedes-benz.com). Domain Name Registrations In a cooperative agreement] 4 with the U.S. National Science Foundation, Network Solutions Incorporated (NSI) of Herndon, Virginia has since 1993 administered the registra- tion of the generic TLDs .corn, .net, .org, .gov and .edu. The .mil domain is managed by the U.S. Department of Defense. The remaining .int, a specialized TLD intended for interna- tional treaty organizations, will, pending final agreement, be administered by the International Telecommunication Union (ITU) in Geneva, Switzerland. As mentioned, the .gov, .edu and .rail domains are restricted to U.S. applicants by allo- © TeleGeography, Inc. 1996 Internet Domain Names cation policy. This leaves .corn, .net, and .org, which should be considered as international TLDs (iTLDs). The NSF and NSI have been struggling to deal with the astronomical growth and commercialization of the Internet. When NSI began domain name registrations in the spring of 1993, approximately 400 domain names were being regis- tered per month. However, in September 1996, this had risen to about 75,000--approximately a 15 percent growth rate per month. Since the overwhelming percentage were in the .com category, the NSF found itself essentially subsidiz- ing commercial registrations which did not fall into its man- date. Reacting to this, NSF amended its cooperative agree- ment with NSI, allowing it to institute a US$50 a year charge for domain name registrations from 14 September 1995.] 5 Box 1. The Internet’s Unelected Governors Who? IETF IESG Where You Can Find Out More Internet Engineering Task Force http://www.ietf.cnri.resto n.va.us/ Internet Engineering Steering Group http://www, ietf,cnri.reston.va,us/iesg.html What You Really Need to Know These are the engineers who make Internet standards. You’re essentially a member of the IETF if you participate in their work or pay to attend one of their tri-annual meetings. Handles the internal management of the IETF. Formed by Area D~rectors who handle Working Groups, an IETF Chair, and an IETF Executive Director who is principally funded by the US Government. lAB Internet Architecture Board The Internet Architecture Board (lAB) is a http://www.iab.org/iab/ technical advisory group of the Internet Society (ISOC). ISOC Internet Society A "non-profit, scientific, educational and http://www.isoc.org/ charitable" entity, incorporated in 1992 in the District of Columbia, USA, separate from the Corporation for National Research Imtiatives (CNRI), to which it formerly belonged. IANA Internet Assigned Numbers Authority http://www.isi.ed u/iana/ See also ISI at http://www.~s~.edu/ An "lnternet Service""of the High-Performance Computing and Commumcations (HPCC) Division of the Information Sciences institute {ISI), part of the University of Southern California’s (USC) School of Engineering. CRNI Corporation for National Research Initiatives The IETF secretariat is located at the htCp://wv~v.cnr~.reston.va.us/ Corporation for National Research Imtiatives (CNRI) NSF National Science Foundation http:flwww.nsf.gov/ See also NCRI at http://www.cise.nsf.gov/n c ri/in d ex.html InterNIC http://wwvv.inte rm c.n et/ The directory service can be found at http://ds.mternic.net/. The registration service is at http://rs.interm c.netJ. InterNIC NSI Network Solutions Incorporated http://www.n etsol.c ore/ See also SAIC at http://www.saic.com/ Provides support and grants for research in networking and commumcations including the now commercial NSFNet. The re]evant group is the Networking and Communications Research and Infrastructure (NCRI) Division in the Directorate for Computer and Information Science and Engineering (CISE). A project compr=sed of two distinct serv=ces partially funded by the NSF: Directory and Databases Services managed by AT&T and Registration Serv=ces managed by NSI. The company that performs most generic domain name registrations in a cooperative agreement with NSF. NSI is a subsidiary of a major privately owned U,S, Department of Defense contractor, Sc=ence Applications International Corporation (SAIC). SAIC has proposed to acquire Bellcore, the R&D company cooperatively owned by the U.S. Regional Bell Operating Compames. J TeleGeography 1996/97 © TeleGeography, Inc. 1996 lnternet Monikers and Money Among lnternet users, the fee led to widespread negative reaction. Though the initial anger has now died down some- what, criticism continues regarding the amount of money being made by NSI, a private corporation with a U.S. gov- ernment blessed monopoly over essentially international name resources. During the period from September 1995 to September 1996, the new annual fee generated on the order of US$55 million in revenue (two years must be paid up front). 16 If the current monthly growth rate (15 percent) continues unabated, the fee will generate more than US$158 million in annual revenue for NSI. It is not clear what the actual costs are of performing a domain name reg- istration. However, a significant revenue stream will accrue from the yearly renewal fee which involves little more than sending out a bill. Of the fees collected, NSI keeps 70 per- cent and the rest goes into an lnternet "Intellectual Infrastructure" fund administered by the NSE As of September 1996, this fund contained around US$7 million. Clearly, Internet domain name registries are now a big busi- ness. Domain names are also being brokered for increasing amounts of money. For example, when Microsoft wanted "slate.corn" for their new VVVVVV-based interactive magazine of politics and culture, they reportedly purchased it for $10,000 (through a third party for obvious reasons). Undoubtedly greater amounts have been paid to obtain strategic domain names. Sometime the price is too high. An offer of US$50,000 is reported to have been rejected for "television.com." When AT&T wanted to launch their new Internet service, they attempted to purchase the domain name "worldnet.net," owned by a Paris-based Internet ser- vice provider. But even they balked at the reported US$500,000 asking price. Box 2. Domain Names and Trademarks The principle for regtstration of domain names has been till now first come, first served. Some domain names relating to famous trademarks like "Tup.com" and "mcdonalds.com" were snagged early on by Internet aficionados and were handed over to their respective companies for token gifts or donations to charity. However, the realization that big companies were often willing to pay thousands of dollars for a mnemonic cyberpresence has lead to many cases of what has come to be known as domain name "hqacking. Many a company has realized after requesting an appropriate domain name that it belongs to somebody else. For instance, "coke cam" currently belongs to Rajeev Arora of California and "rolex.com" is registered to Jamce Ard of Colorado. At least In the case of well known trademarks (and especJally when faced with legal action), most amateur speculators can be bribed to amicably relinquish an important name. Caught in the middle between domain name holders and trademark owners, NSI lnsbtuted a policy m July 1995, later modified in November 1995, that gave trademark owners the ability to reclaim a domain name if they could produce a corresponding federal (national) trademark. But this policy has neither mollified trademark owners or domain name holders Many rightly point out that trademark law allows multiple entities to share the same name if they are not in competing businesses and/or the same geographical area--for example, there can be a McDonald’s computer company and a McDonald’s hamburger company. However, on the Internet, there can only be one "mcdonalds.com" (now owned by the latter). Trademark lawyers also discovered that a trademark from any country could possibly be used to defend the usage of a domain name. This led to trademark registrations by domain name holders in places like Tunisia, where in 48 hours, a "federal" trademark can be produced. The NSI policy also does not deal with domain names that, although not identical to a trademark, can constitute infringement of U.S. "common law" trademarks not registered under the Lanham Act One of the basic problems that NSl faces Js that it is handing out international domain names but that existing trademark law is fundamentally national--there is no such thing as a widely recognized "international trademark. In the non-lnternet world, there is an established tradition of dealing with trademarks on a national basis. Trademark rules may be different in the U.S., France or Japan but in each country there is a procedure and clear jurisprudence for dispute resolution. This is strongest at the national level with occasional provisions at the regional and international level Trademarks are only the first of many issues that will fundamentally put territorial based legal systems into conflict with the Internet. The relationship between trademarks and domain names has been described in detail elsewhere. One of the best summaries is probably David Maher’s "Trademarks on the lnternet. Who’s tn Charge?"(http://www.aldea.com/clx/make.html). Another survey is "What’s In a Name?" from students at the Georgetown University Law School (see http://www law.georgetown.edu/lc/intermc/doma~n 1.html). © TeleGeography, Inc. !996 Internet Domain Names F International or National Top Level Domains? Arguments for New International TLDsI, Arguments for Country-Code-based TLDs ¯ Respects national sovereign~ in definition of National Information Infrastructure (Nil) policies ¯ Already over 400,000 registrations in .corn, .net and .org exist that would be impossible to "repatriate" to count~/-code-based TLDs ¯ Marketplace seems to have ~ndicated that domain names without country codes are preferred ¯ Globalization trend in telecommunications services (e.g., international freephone services) ¯ Introduces competition in domain name reg=stries (breaks NSI monopoly) ¯ Facilitates alternative access to what has become a "scarce" resource (.corn) ¯ Respects national naming conventions (e.g., co.uk, plc.uk, ltd.uk) and national language conventions (e.g., .gov.uk, gouv.fr) ¯ Provides an up-front national registry-level basis for rejecting name registrations with potential trademark conflicts ¯ Provides a basis for dealing with domain names that are considered offensive ¯ International TLDs further complicate the legal quagmire of trademark issues dispute resolution ¯ New international TLDs would lead to more users who acquire "rights" in internabonal name space without a legal framework J Evolving Domain Name Space Proposals for the evolution of DNS have gained momentum since late 1995 with a series of workshops sponsored by the Harvard Information Infrastructure Project, the NSF, the Commercial Internet exchange (CIX) and the lnternet Society (ISOC). 17 In November 1995, an Internet draft was published by B. Carpenter et al., "Proposal for an ISOC Role in DNS Name Space Management." It suggested that the "lnternet Society take a formal role in the oversight and licensing of compet- itive registries for international Internet name space." This was followed by another Internet draft in January 1996 by R. Bush et al., "Delegation of International Top Level Domains (iTLDs)." Most recently, Jon Postel, Head of the lnternet Assigned Numbers Authority (IANA) has authored a proposal "New Registries and the Delegation of International Top Level Domains"18 that plans the creation of up to 150 new international top level domains (like .com, .net and .org). Up to 50 new registries performing functions similar to NSI would be awarded. In this scheme, ISOC would be the "international, legal and financial umbrella" of IANA. There are both pros and cons to creating new international top level domains (see Box 3). While doing so may or may not be a good idea, the problems here seem really of a process nature. Which parties should be consulted in reforming of international Internet name space? Certainly many more than are currently involved. Many of the poten- tially concerned parties such as the WWW Consortium, the European Commission DGXIII, R~seaux IP Europ~ens (RIPE), and Asia-Pacific Network Information Center (APNIC), have been listed by Anthony M. Rutkowski, ex-Executive Director of ISOC, in a recent paper posted on the Internet.19 It is also obvious that many governments also want to have a say in issues related to a nascent Global Information Infrastructure. As Rutkowski as pointed out, even though any DNS policy decision would have an impact on interna- tional name space, no non-U.S, public policy officials have ever been involved in the public meetings held to date. Would a neutral solution for international name space real- ly be encouraged by bringing the ISOC into the process? Reading between the lines, it’s hard to see. According to Postel’s draft, these potentially multi-million dollar generat- ing registries will be awarded by an "ad-hoc working group" comprised of three people from IANA, two from the Internet Engineering Task Force (IETF) and two from ISOC. Yet these groups of engineers (albeit well intentioned) have a very limited legal or policy authorization. For ISOC (or any group) to claim responsibility for delegating international name space when there are such high stakes appears unsus- tainable. The Internet has become far too commercial and strategically important as a global communications tool to simply perpetuate the same informal arrangements that have kept it glued together until now. Interestingly, ISOC has now seemed to recognize this and has proposed that iTLD policy be set by an "International Ad Hoc Committee" (IAHC), including a more international representation. At the recent Montreal IETF/INET meeting in June 1996, however, the ISOC Board of Trustees voted, in principle, to back the new proposal. Note that the last version of the pro- TeleGeography 1996/97 © TeleGeography, Inc. 1996 posal calls for ISOC to receive 2 percent of the annual income from the registries. How DNS Rules Affect The Internet’s Workload Historically, the servers responsible for the reot (sometimes referred to as the "." zone) have also been responsible for domain name lookup services for all international top level domains (iTLDs) like .corn, .net, and .org. With one excep- tion, these servers are all located in the United States. They are operated by Network Solutions in Virginia, the Information Sciences Institute (ISI) at the University of Southern California, large lnternet service providers, NASA, the University of Maryland, Paul Vixie (primary author of DNS code) in California, two by the U.S. Department of Defense, and one by NORDUnet in Sweden. The IANA has traditionally had responsibility for coordination of these sites. There has been some discussion of diversifying the geo- graphical location of the root servers and placing them more towards the "core" of the global Internet; starting with experimental root servers in Asia and Europe (DNS technol- ogy allows up to 13 root servers). These still inconclusive deliberations are taking place in the the lnternet Engineering and Planning Group (IEPG); a group mainly comprised of lntemet Service Operators.2° There are also grumblings in the Internet technical commu- nity that, considering the amounts of money that NSI takes in from .com registration services, it should more directly contribute to the costs of running the root name servers. And in fact, NS1 has offered machinery and funding to these groups. However, the uneasy tension between NSI and the Net technical community (which feels NSI has been unfairly handed a cash cow by NSF) has resulted in some of these offers being rejected to maintain independence. At the root level, the limited number of new international top level domains being proposed would have a minimal impact on the root servers since it is hard to imagine that there would ever be more than 500 entries (there are cur- rently about 200 ISO 3166 TLD entries plus the existing iTLDs). Therefore, the service requirements for the root level are several orders of magnitude less than lookup services required for the current 550,000 entries in ".com." However, the fact that the root server are run for free, com- bined with the increasing load of existing iTLDs (e.g., .corn) means it may make sense to remove iTLD name services from the root servers. Technically there is no reason that this couldn’t happen. In fact, considering the somewhat different operational and administrative requirements of the root and iTLD name ser- vices, it is probably a good idea. One catalyst might be new international top level domains. There is the sentiment that the considerable income garnered from domain name regis- tration services should be used to directly pay for distrib- uted iTLD name servers. And since the plan for new iTLDs is that they be awarded to new registries around the globe (where related name servers would be co-located), the eventual popularity (which is unclear) of rn_D name servers than the U.S.-centfic model that exists today. Conclusion The Internet is in a painful transition period and appears to be caught in a cross-tire between tremendous commercial, political, legal and operational interests and anarchistic Net individuals who still want to do their own thing. While chaos may still be one of the Internet’s strengths, others are con- cerned with a certain stability of infrastructure, internation- al comity, respect for legal issues such as trademarks and accountability which is part of the real world. The current conflicts concerning DNS are likely only the tip of an iceberg. What seems important to the Internet techni- cal community will need to be increasingly balanced with the views of other communities--an inescapable consequence of the new strategic and commercial importance of the Net. In fact, the skirmishes within the trademark community may well prove to be minor when compared to likely future bat- tlegrounds where the stakes are higher (e.g., lnternet tele- phony and interexchange settlements for backbone providers). In this new Internet of many stakeholders, it’s going to get increasingly difficult to reach that IETF credo of "rough consensus." ¯ © TeleGeography, Inc. 1996 Internet Domain Names Endnotes 1. See MIDS Press Release: Global Domain Names Grow Rapidly Worldwide at http://www.mids.org/prdomreg.html. 2. The ’Request for Comments’ (RFC) document series is the official publication channel for Internet standards documents and other publications of the IESG (lnternet Engineering Steering Group), lAB (Intemet Architecture Board), and Internet community. See C. Huitema et al., "Not All RFCs are Standards," RFC 1796, ftp://ds.internic.net/rfc/rfc1796.txt, April 1995. Any RFC can be retrieved with the address ftp://ds.internic.net/rfc/rfcxxx.tx~ where xxx is the RFC number. 3. D.L. Mills, Internet Name Domains, RFC 799, ftp://ds.internic.net/rfc/rfc799.txt, September 1981. 4. J. Postel, J. Reynolds, Domain Requirements, RFC 920, ftp://ds.internic.net/rfc/rfc920.txt, October 1984. 5. ISO 3166, see http://www.iso.ch/cate/d22748.html. The standard defines two character (alpha-2) and three character (alpha-3) versions. The Internet domain name system uses the alpha-2 version. 6. Notably the Asian-Pacific Network Information Center (APNIC) at http://www.apnic.net and R~seaux IP Europ~ens (RIPE) at http://www.ripe.net. 7. See 5e REUNION DU COMITE DE CONCERTATION NIC FRANCE du 14/02/96, http://www.nic.fr/Presentation/CC/cc-960214.html. 8. See notes from workshop Internet Names, Numbers, and Beyond: Issues in the Coordination, Privatization and Internationalization of the lnternet, http://ksgwww.harvard.edu/iip/nsfm~nl .html, November 20, 1995. 9. See http://www.nic.uk/press.htmt and http://www.domain-registry.nl/. I0. See http://www.nic.fr/Presentation/GT/gt-960320.html. 11. See http://www.nic.ch. 12. See Domain Names within the U.K. at http://www.nic.uk/new/domains.html and http://www.nic.uk/proposals/rupl .html. See also Pages Jaunes des Domaines de la Zone Francalse (FR) at http://www.nic.fr/Annuaire/index.html. 13. A. Cooper and Jon Postel, The U.S Domain, RFC 1480, ftp://ds.internic net/rfc/rfc1480.txt, June 1993. 14. See NSF Cooperative Agreement at http://rs.internic.net/nsf/agreement. 15. See http://www.~soc.org/adopsec/nsf-name~fees.html and http://rs.~nternic.net/announcements/index.html. 16. See NSF Cooperative Agreement No. NCR-9218742, Amendment 4 at http://rs.lnternic.net/nsf/agreement/amendment4.html. 17. Internet Names, Numbers, and Beyond: Issues in the Coordination, Privatization and Internationalization of the Internet, convened in November 1995 in Washington, D.C. The workshop was sponsored by NSF’s Networking and Communications Research and Infrastructure (NCRI) Division and the Harvard Information Infrastructure Project (liP). A second workshop on Internet Administrative Infrastructure was held m February 1996 sponsored by the Commercial Internet Exchange (CIX) and ISOC. A third related workshop Coordination and Admimstration of the Internet was held in 1996 in Cambridge, Massachusetts, sponsored by the Harvard liP, CIX, ITU and ISOC. 18. Jon Postel, "New Registries and the Delegation of International Top Level Domains," Internet Draft, ftp://ds.internic.net/Intemet- drafts/draft-postel-iana-itld-admin 01 .txt, June 1996. 19, See http://www.wia.org/pub/policy-orgs.html. See also Richard Hovey and Scott Bradner, "The Organizations Involved with IETF Standards Process," Internet Draft, ftp://ietf.cnri.rest~n~va.us/internet~drafts/draft-~etf-p~ised95-ietf-~rgs~3.txt~ June 1996. 20. For further discussion on how DNS can geographically spread the Internet’s workload, see T~ Bnsco, "DNS SupportlforLoad Balancing," RFC 1794, ftp://ds.internic.net/rfc/rfc1794.txt, April 1995. TeleGeography 1996/97 © TeleGeography, Inc. 1996 © TeleGeographyo Inc. 1996 Mapping the lnternet MAPPING THE INTERNET How does the Internet work? A map of the infrastructure--the main backbone networks and Network Exchanges~oes not tell you very much about how packets of data actually get from one computer to another. Likewise, if you simply map the flow of packets from one site to another you can easily miss how the money flows. Hence, the following pages try to provide an overview of how the Internet works by presenting three differ- ent views: commercial, operational and geographical. Each provides part of the answer. F Internet Cash Flows Network Exchanles~ and Other Cash flows on the Internet begin with the end user (e.g., an individual, company, or university) that pays an Internet service provider (ISP) for access. Many small ISPs, in turn, pay larger ISPs for access to their networks. Each ISP must directly or indirectly connect with, and pay for access to, a Network Service Provider (NSP). The NSPs consist of regional, national and international backbone providers that connectto each other at Network Exchange points, tn North America these Exchanges are commonly known as Network Access Points (NAPs) or Metropolitan Area Exchanges (MAEs). The major North American Exchanges are profiled at p. 52 below. In some cases the functions of ISPs and NSPs are consolidated into a single entity (e.g., Internet MCI). Note: The chart ~s ~ntended to be illustrative of economic relationships in the U S only. The U S Federal Government, which originally subs~d{zed the net~vork for scientific and defense purposes, has w~thdrawn most funding but continues to contribute towards research and essential services. ©Tele6eography, Inc. 1996 TeleGeography 1996/97 © TeleGeographyo Inc. 1996 Figure 2. Internet Packet Flows: End Users to Global Backbones Competition and rapid growth are bringing both diversifica- tion and stratification in Internet infrastructure. Global con- nectivity can be assured only through a complex set of peering and transit relationships between providers. Peer networks exchange their own customer routes with each other; a network providing transit to another allows its backbone to be used to reach a destination not on its own network. Individuals or enterprise networks choose providers based on service offerings, cost, support, and quality of service. KEY --- = customer ortransit relationship = peering relationship 6BP = global backbone providers IBP = international backbone providers NSP = national service provider RSP = regional service provider ISP = internet service provider/reseller End User = individual or enterprise network © Dimension Enterprises, Inc. 1996 © TeleGeographyo Inc. 1996 Mapping the lnternet anges: Where the Backbones Meet Until 1994, the backbone transmission facilities used for the Internet were underwritten by the U.S. National Science Foundation (NSF) As multiple commercial backbones took over the government’s role, the NSF established a series of pri- ority network access points (NAPs) for national and regional networks to interconnect. The high priority NAPs designated by the NSF included: Sprint’s New York NAP, Amentech’s Chicago NAP, and PacBell’s San Francisco NAP. In addition, the MFS metropoli- tan area exchange (MAE) for Washington, D.C., MAE-East, was given special priority status. A similar facility in northern California, MAE-West, also attracted a number of networks The NSF’s system of NAPs could not keep up with the growth of new backbones and trafhc, however. Packets destined for an e-mail box located across the street in Los Angeles, for example, might take a trans-continental journey to MAE-East and back for delivery. Bg early I996 mang network service providers began to establish regional exchanges that could avoid the congestion at the major U.S. national exchanges (see Table 1 at p. 52). These new sub-national switching points for Internet traffic pro- vide a low-cost alternative for local interconnection to the NSF priority NAPs, other national exchanges, and continental tran- sit backbones. In the simplest designs, Internet service providers (ISPs) link to an exchange using frame relay protocol or a shared Ethernet hub. As the exchange trafhc increases, other technologies are easily/ntroduced. Most of these facil~- ties are located in large metropolitan areas in which a s/gn/h~ cant portion of the trafhc (estimated between I 0 to 50 per- cent), is expected to remain local. The regional exchanges may prove effective in reducing traffic routed over the more expensive large exchanges Further, as major national transit providers begin moving towards private interconnects, the balance of trafhc routed through private, national and regional interconnects can be expected to change dramatically in 1997. --Barbara Dooley Barbara Dooley, bdooley@cix.org, is Executive Director of the Commercial Internet exchange Association (CIX), former managing edi- tor of ClXtra, its monthly newsletter, and senior consultant with Dimension Enterprises, Inc, a Herndon, VA internetworkmg consulting firm, TeleGeography 1996/97 © TeleGeography, Inc. 1996 The Internet’s Big Switches: Major U.S. Network Access Points (NAPs) City J NAP Name i Type of J Facilities Owner Networks Connection j or Operator Connected Region Eastern Massachusetts New Jersey Boston Boston MXP Pennsauken NY NAP (NSF) F X-East New York (Manhattan) NYIIX New York NYRIX Manhattan) New York (Manhattan) I MAE-NY Tysons Corner, VA MAE-East ~rner, VA I FIX-East Imovlng to Pennsauken) l New York NAP Type Regional Nat=onal Internatmnal Federal Nat=onel ;Internatmnal Regional Regional Washington, DC Nat=onal Metro I lnternat~onal Federal Frame Shared Ethernet Switched Ethernet FDDI Sw=tched Ethernet FDDI Switched Ethernet Sw=tched Ethernet FDDI Sw=tched Ethernet FDDI Swn~ched Etharnet Switched FDDI FDDI Swrtched FDDI Switched Et hernet Management Analys=s Inc. (MAI) 21+ 6 pending Sprint ; Telehouse America 12 +4 pending I Telehouse America [planned MFS Datanet i planned MFS Datanet !55+ US Government i3 Central ,lhno,s !Chicago ! J Chicago NAP (NSF) National ATM FDDI IAmeritech (AADS) /18+2 pending : Ch,cago j MAE-Ch cago Reg,onal FDDI I MFS Oatanet 13 I Switched Ethernet Michigan Detroit Detroit MXP Regional FOOl I MAI !2 Switched Ethernet Shared Ethernet TCNS M,ssourl i St Louis I STLOUlX Regional Shared Ethernet i Data Research 12 I Southern Texas ’,Dallas I MAE-Oallas Reg,onal ’, Sw,tched ethernet I MFS Datanet n.a. i Houston ’ MAE-Houston Regional I Switched Ethernet I MFS Datanet 4+ pending___._ Western Arizona Cahforma Phoemx I Phoenix REP o!~lonal i Switched Ethernet Tucson I Tucson NAP Regional Frame Relay i ____ Switched Et hernet Tucson ITucson Interconnect FrameRe]ay l San Francisco I SF NAP (NSF) National I FDDI I Internat~onat ATM Paid Alto ! D=g~tal Internet Reg=onal ’FDDI I Exchange PaiD Alto National I (PAIX) nternat~ona Paid Alto/Santa Clara ; CIX NAP National FDDI (mowng to Paid Alto) nternatmna SMDSFrame Relay I MAE-West Nat=onal Internat=onal San Jose Mountain View I Los Angeles I Los Angeles Los Anqeles Utah i Salt Lake City Northwest ’. Oregon/Washington MAE-West FiX-West FOOl Switched FOOl ! Sw=tched Ethernet FDDI ’ Sw=tched FDDI .. Switched Ethernet ATM Switched FDDI i Sw=tched Ethernet ! Switched Ethernet i Frame Relay Shared Ethernet Switched Ethernet Frame Relay I LA NAP I MAE-LA I Utah REP NW Internet eXchange ( I RTD 15 RTD 14+ pend=ng ACES Research 18 Pacific Bell ]22+ pending Digztal Equipment ~6 pend=ng Corp D=gital Equipment ~13SMDS Corp./~Ndtel 117 direct connect (transltmn i pend=ng to D~gital) MFS Datanet 137 pending NASA-Ames Pacific Bell 11 Universlty of 13 Southern Cahforma I MFS Datanet I planned Inquo ]4 EL! National Federal Regional Regmnal Regional Reg=onal Regional Acronyms: ATM = Asynchronous Transfer Mode FDDI = Fiber Distributed Data Interexchange SMDS = Sw=tched Mult~megab=t Data Service TCNS = Thomas Conrad Network Serv=ce ©Commercial Internet eXchange Association 1996 © TeleGeographyo Inc. 1996 Mapping the Internet F High-Speed Internet Backbones, 1995-96 Like a river system or tree canopy, the Internet com- prises millions of tiny connections feeding into progres- sively larger links that ultimately connect to a few large trunks. In the case of the Internet, these "backbones" are leased circuits on fiber-optic cables connecting switching computers called routers. The size of the backbones is constantly growing. In 199 I, the American NSFNet backbone consisted of T- I (I .5 Mbps) lines, which were the highest-capacity lines used for the Internet at that time. In the early 1990s, this backbone was upgraded to T-3 speeds (45 Mbps), and now several commercial T-3 networks cover the United States. Even 45 Mbps may soon seem slow, as MCI has installed 0C-3 (155 Mbps) lines on both its own U.S. network and the vBNS network it operates for the National Science Foundation. And in June it announced plans to upgrade its commercial network to OC- 12 (622 Mbps). Meanwhile, the new availability of leased E-3 (34 Mbps) circuits in Europe has inspired plans for equiva- lent European backbones. Such upgrades would reduce the enduring American focus of the Internet. Instead of being routed from one European country to another via New York or Washington, packets may soon travel via Paris or Stockholm. Asian providers have similar dreams, but even a T-l backbone in Asia (such as those planned by Asia lnternet Holding and Singnet) would be an improvement over the sparse intra-Asian links that now exist (Japan, Singapore and Hong Kong are linked by 2 Mbps lines). A recent initiative to improve bandwidth is the Internet Railroad, part of the Internet 1996 World Exposition (http://park.org). The Railroad’s stated goal is "building a backbone circling the world at 45 million bits per sec- ond. As of July 1996, only a few long-distance links were in operation, but the Tokgo-Seoul link (operated by KDD and Korea Telecom) alone is by far the largest international link within Asia. It is not clear whether the Internet Railroad links will remain in operation after the end of the Exposition in December 1996. The maps on the following pages show all Internet backbones of 34 Mbps speed or greater for which infor- mation is publicly available. Planned backbone expan- sion in the United States was not shown to avoid clut~ tering the map with dozens more links. --Zachary M. Schrag TeleGeography 1996/97 © TeleGeographU, Inc. 1996 © TeleGeography, Inc. 19~6 Internet Backbones Dante Europanet (March96} Ebone (Junsss! UUnet Pipex (~ Julyg6) Bashed lines indicate planned links. [] ~xchaoge/Access Points All links are 34 Mbps (E-3), Each network is shown as it existed orwes planned on the date given in parentheses. Squares represent metropolitan areas and include nearby cities. EUnat and Global One also have a trans-European network but have anneanced ns plans to upgrade to 34 Mbps. © TeleGeography, Inc., 1996 Mapping the Intemet Asia 55 TeleGeography1996/97 © TeleGeography, Inc. 1996 Figure 3. Traffic to and from the U.S. via FIX-West, July 1996 Not ldenefied Japan Canada Hong Kong Netherlands Umted Kingdom New Zealand Denmark Talwan Korea Germany Poland Italy France Sweden Malay~,a Spare Argenena Swlzerland Panama Turkey Porlugal Guam Israel Albama Greece Colombia Llthuama Thadand Venezuela Costa Rice Iceland Sent to U.S. domains Bytes Received from U.S. domains 100,000,000 1,000,000 10,000 100 0 100 10,000 1,000,000 lO0,000,00o Source. FIX-West stalsbcs data summaries, http’//WW~N.nlanr.netlNA/FIX/Stats/West! © TeleGeography, Inc. 1996 © TeleGeographyo Inc. 1996 Mapping the lnternet Table 2. Host Computer Growth by Domain, 1995-1996 Domain Hosts, July 1995 Hosts, July 1996 Growth, 95-96 Commercial (corn) 1,743,390 3,323,647 191% Education (edu) 1,411,013 2,1 !4,851 150% Network (net) 300,481 1,232,902 410% United Kingdom (uk) 291,258 579,492 199% Germany (de) 350,707 548,168 156% Japan (jp) 159,776 496,427 311% United States (us) 113,226 432,727 382% Military (rail) 224,778 431,939 192% Canada (ca) 262,644 424,356 162% Australia (au) 207,426 397,460 192% Government (gov) 273,855 361,065 132% Organization (org) 201,905 327,148 162% Finland (fi) 111,861 277,207 248% Netherlands (nl) 135,462 214,704 158% France (fr) 113,974 189,786 167% Sweden (se) 106,725 186,312 175% Norway (no) 66,608 120,780 181% Italy (it) 46,143 113,776 247% Switzerland (ch) 63,795 102,691 161% South Africa (za) 41,329 83,349 202% New Zealand (nz) 43,863 77,886 178% Denmark (dk) 36,964 76,955 208% Austria (at) 40,696 71,090 175% Spain (es) 39,919 62,447 156% Korea, Republic of (kr) 23,791 47,973 202% Note: The vast majority of computers using three-letter domains (com, edu, org, gov, rod, and netl are located m necessarily located m the same country as the domain name. Source Network Wizards, Internet Domain Survey (http://www nw.com) the United States Hast computers are not Want to Keep up on Mapping Cyberspace? Visit Martin Dodge, a researcher at the University College of London and keeper of the "Geography of Cyberspace" homepage. His Web page has extensive links to sites and resources which address everything from network topology to Web demographics (http://www.geog.ucLac.uk/casa/martin/geography_of_cyberspace.html). A country-by-country listing of international network connectivity (i.e., who has full Internet access and who has e-mail only) is maintained by Larry Landweber, Computer Sciences Department, University of Wisconsin and can be obtained via anonymous ftp from ftp. cs. wisc. edu in the connectivity_table directory. Maps are also available. TeleGeography1996/97 © TeleGeography, Inc. 1996 International Facilities and Carriers TeleGeo~aphy 1996/97 © TeleGeogr~phg, Inc.. 1995 Trans-Pacific Cable Systems Year in Cable Cost (US$) Capac.ity Service System per voice path (voice paths) 1957 Hawaii 1" 378,000 91 1964 TPC-I* 406,000 167 1974 Hawaii 2* 41,000 1,690 1975 TPC-2* 73,000 1,690 1988 TPC-3 16,000 37,800 1991 North Pacific Cable 5,000 85,000 1992 TPC-4 5,500 75,600 1996 TPC-5 2,000 605,000 1997 FLAG 1,500 605,000 Visit our web site at http:HvWnN.telegeography.com for more telecommunications maps. ©TeleGeography, Inc. 1996 © Tele~eograpl~yo Inc. 1996 Submarine Cables MAJOR SUBMARINE CABLES Trans-Atlantic Cable Systems Year in Cable Cost (US$) Capac.ily Service System per voice path (voice paths) 1956 TAT-l* 557,000 89 1965 TAT-4* 365,000 138 1970 TAT-5* 49;000 1,440 1983 TAT-7* 23,000 8,400 1988 TAT-8 9,000 37,800 1989 PTAT 6,000 85,000 1993 TAT-10 2,700 113,400 1994 CANTAT-3 1,000 302,000 1996-97 TAT-12/13 1,000 600,000 UNISUR pital and construction costs only, stated in US$ to the nearest $500, unadjusted for inflation. Current technology permits approximately 5 virtual voice ~aths to be derived from a digital channel operating at 64,000 bits per second (64 Kbp$). Fiber optic cables are expected to have a useful life of at least 25 years. Table reports average cost per voice path for cables with multiple landing points. For example, the TAT-9 system connects the U.S. and Canada with the U.K., France and Spain. The average U,S.-U.K. cost per voice path is approximately $4000. Reserve capacity of cables is generally excluded. Source: FCC and carriers. ©TeleGeography, Inc. 1996 TeleGeography 1996/97 © TeleGeography, Inc. 1996 Route System Lead Investors/Owners Capacity Trans-Atlantic Atlantic Express 1 and 2 Atlantic Express Co. 10 Gbps (private U.S. company) MFS 1 and 2 MFS Communications, 10 Gbps (Gemini) Cable & Wireless KDD and AT&T Corp. Trans-Pacific Europe-Middle East- Asia TPC-6 =100 Gbps FLAG NYNEX, Marubeni, Gulf Assoc., 10 Gbps Dallah-AI-Baraka Group SEA-ME-WE3 France T~l~com, Singapore 10 Gbps Telecom, KDD and 75 others Trans-Siberian Link Rostelecom and other 1.2 Gbps European carriers Trans-Asia-Europe (TAE) Deutsche Telekom, 560 Kbps- Turk Telekom and 1 Gbps MPT China Service Date 1998-99 1998-99 2000-2001 1997-98 1998-1999 1996-97 1997 Africa-Middle East- Africa One AT&T and other 10 Gbps 1999-2000 Asia African carriers South Africa-Far East Telkom South Africa and 10 Gbps 1999 (SAFE) Telekom Malaysia Notes: This table only lists international cable systems which have been partly butlt or have the greatest probabihty of being constructed. The very large capac=ty of fiber optic cable systems now being planned means that only 1 or 2 new systems are hkely to be installed on any g~ven route dunng the next 5 to 7 years. As well, many new systems (TPC-5; MFS-1 and 2) have a "ring" topology so that, m the event of a cable break, traffic can be rerouted over the undamaged path. ©TeleGeography, Inc. 1996 Table 2. Cable and Satellite Capacity on Trans-Atlantic and Trans-Pacific Routes, 1986-2000 Trans-Atlantic (North America-Europe) Trans-Pacific (North America-East Asia) Voice Paths Voice Paths Year Cable 1986 22,000 1987 22,000 1988 60,000 1989 145,000 1990 145,000 1991 221,000 1992 296,600 1993 410,000 1994 701,800 1995 1,310,800 1996 1,310,800 1997-2000" 1,310,800 *minimum available Satellite 78,000 78,000 78,000 93,000 283 000 283 000 496 000 620 800 620 800 710 800 710 800 737 500 Cable Satellite 2,000 39,000 37,800 39,000 37,800 39,000 37,800 39,000 37,800 39,000 114,200 27,000 190,500 27,000 264,000 83,300 264,000 234,000 264,000 234,000 864,600 234,000 1,464,600 424,500 Notes: Estimates of cable vo=ce paths assume that 5 wrtual voice paths can be derived from one 64 kbit/s d~gltal c=rcu=t; cable est=mates do not ~nclude c~rcu=ts held in reserve for cable/satelbte restoration services. Capacity estimates exclude proposed systems hsted ~n Table t above. Estimates of trans-Pacific cable c=rcu~ts are based on capacity from North America to Japan via Hawah or Guam. Estimates of satellite voice paths are based on Intelsat satell=tes only prior to 1993; satellite est=mates exclude one tntelsat satellite in each region held in reserve. Estimates also assume one vmce path per channel until 1989 deployment of Intelsat VI series with 24,000 channels or 120,000 voice paths using Dig=tal Code Multiplication Equipment (DCME). The Intelsat VII series, deployed m t992, has a nominal capacity of 18,000 channels or 90,000 vmce paths using DCME. For 1993-2000 time period, estimates assume full capacity of the following non-lntelsat systems is available: PAS-1, PAS-3, Ormn-1 and TDRS-4 (trans-Atlantic); and PAS-2 Rimsat/Express (2 satellites) and TDRS-174 (trans-Pacific). Source: FCC and carriers ©TeleGeography, Inc. 1996 © TeleGeography, Inc. 1996 Submarine Cables Box 1. The Next Generation of Cables: 100 Gbps and Beyond The third generation of undersea fiber optic cables now entenng service (TAT 12/13; TPC 5) can carry approximately 5 Gigabits per second (Gbps) per fiber pair or approximately 320,000 virtual voice channels. This represents an order of magnitude increase from the second generation of cables (operating at 560 Mbps) which, in turn, provided a tenfold increase in capacity over first generation cables such as TAT-8. Recent trials and experiments by AT&T, AJcatel and KDD suggest that the next generation of cables, to be deployed in the 2000- 2005 timeframe, will increase capacity by at least another order of magnitude to 50 Gbps and probably to 100 Gbps or more. That will be enough to transmit at least 3.5 million simultaneous telephone calls or several hundred thousand channels of com- pressed video services. The enormous capacity of the next generation of fiber optic cables will result from two new technologies--optical soliton transmission and wave division multiplexing (WDM)--which leverage the bene- fits of earlier breakthroughs, such as optical amplifiers. Digital communications generally are sent over a fiber optic cable by very rapidly transforming the original electrical signal into tiny pulses of laser light; the presence or absence of a pulse in a given pedod represents a binary I or 0. However, optical fibers can only carry a signal for a few hundred kilometers before it becomes too blurred or weak to be useable. Thus, long distance fiber optic cables contain repeaters, spaced at regular intervals, to amplif~ the signal. For many years the only way to regenerate a signal in a tong haul cable was to use an opto-electronic amplifier which converted the weak light pulses into an electronic signal, boosted the s~gnal through an amplifier, and then transformed the boosted signal back into light pulses. In the late 1980s, however, amplifiers were developed to regenerate the optical signal without any electronic intermediary. These optical amplifiers typically consist of a few meters of erbium-doped fiber (EDF) inserted into the transmission path and hence are known as EDF Amplifiers or EDFAs. An EDFA permits a signal to be "pumped" up using a laser light source thousands of kilometers away at one of the cable head ends. Notwithstanding optical amplifiers, the bit rate of tong haul cable systems has generally been limited to 5 Gbps due to the way in which the light pulses propagate. But scientists have now devel- oped a way to create unique pulses of light, known as solitons, which maintain their shape and intenszty at very high bit rates over great dzstances. For example, in 1995 KDD demonstrated the feasibility of transmitting a 20 Gbps optical soliton data stream by time division multiplexing 10 Gbps pulses on a 8100 kilometer fiber optic cable test bed. By coupling soliton technology with wave division multiplexing (WDM) the aggregate transmission capacity of any given fiber optic cable may be increased severatfold. In one experiment by Alcatel, sixteen 2.5 Gbps channels, each with a different wave- length, were multiplexed together to create a 40 Gbps data stream over a distance of over 1,400 kilometers. And in February 1996, KDD and AT&T announced they had transmitted over 110 Gbps on a 730 km test bed cable and would deploy WDM technology on a new trans-Pacific cable to be completed by 2001. Later KDD announced that it would use the same technology for a 100 Gbps cable around the islands of Japan. Field trials of WDM technologies elsewhere are also promising: Alcatel has reported WDM transmission of four 2.5 Gbps data streams over 3,500 kilometers on the RIOJA cable system between the U.K. and Spain; AT&T has conducted a similar trial transmitting 10 Gbps over a segment of the Columbus-2 cable between Florida and St. Thomas m the Caribbean. The commercial impact of these developments will be felt well before the next generation of cables is in the water. As shown by the RIOJA and Columbus-2 trials, WDB technologies will permit some cable owners to upgrade capacity merely by changing the equipment at the cable head ends. Four or even eightfold capac- ity increases may be possible. Second, development of WDM techniques is likely to make fiber optic systems increasingly flexi- ble and hence attractive to new investors. For example, because WDM can be used to create different virtual (frequency specific) channels, a cable can be partitioned to satis~ the routing require- ments (landing points) of particular carders or countries without reducing the cable’s overall capacity. Finally, as soliton WDM technology moves into commercial pro- duction, the historical relationship between inter-continental and local prices ~s likely to flip flop. By 2002, for example, a call from Los Angeles to Tokyo via TPC-6 may cost less than a call from one of Los Angeles’ many area codes to another. This is the new tete- com economics which light wave technology w~ll soon usher m. Sources: Franklin W. Kerfoot and Peter K. Runge, "Future Directions For Undersea Communications," AT&T Technical Journal (January/February 1995) Vol. 7z~ #1, pp. 93-100. S.S. Sian, S.M Webb, K.M. Gill, "Sixteen x 2.5 Gbps WDM Unrepeater Transmission Over 427 km," Alcatel Submarine Networks, London (June 1995). H. Taga, N. Takeda, K. Imam, S. Yamamotu and S. Akiba, "110 Gbps [22 5 Gbps), 9500kin Transmission Experiment Using 980nm Pump EDFA 1R Repeaters without Forward Error Correction," KDD Laboratories, Japan (February 1996). Linn E Mollenauer, "Recent Advances in Ultra Long Distance, High Bit Rate Soliton Transmission," Speakers" Papers, 7th World Telecommunicabon Forum, ITU Technology SummJt, vol. 1 (Geneva: ITU, 1995), pp. 761-65. TeleGeography 1996/97 © TeleGeography, Inc. 1996 Table 3. The Top 40 International Carriers, 1990-1995 Outgoing Traffic (millions of MiTTs) Rank Company Country 1995 1994 Change94-95 1993 1990 1 AT&T (a,b) United States 8482 7947 6.7% 7129 6080 2 Deutsche Telekom (c) 6ermany 5244 5147 1.9% 4680 3146 3 MCI (a,b) United States 4452 3517 26.6% 2839 1184 4 BT (d) United Kingdom 2909 2489 16.9% 2310 2170 5 France T~l~com (a) France 2805 2603 7.8% 2576 2126 6 Telecom Italia (e) Italy 1908 1708 11.7% 1610 1045 7 Swiss P-El" Switzerland 1778 1649 7.8% 1572 1356 8 Sprint (a,b) United States 1765 1471 20.0% 1175 577 9 Hongkong Telecom (a,d,f) Hong Kong 1692 1578 7.2% 1377 1120 10 Stentor (b,g) Canada 1467 1525 -3.8% 1552 1344 11 KPN (a) Netherlands 1459 1346 8.4% 1238 905 12 China MPT (f) China 1390 1170 18.8% 900 350 13 Mercury (d) United Kingdom 1107 1018 8.7% 820 354 14 Belgacom (a) Belgium 1106 1049 5.4% 979 731 15 KDD (d) Japan 1079 1011 6.7% 952 764 16 Telef6nica Spain 1025 948 8.1% 847 611 17 Telmex (a) Mexico 950 844 12.6% 825 421 18 Austrian P’l-I(a) Austria 901 819 10.0% 767 559 19 T~l~globe (a) Canada 898 861 4.3% 808 565 20 Telstra (h) Australia 807 690 17.0% 640 565 21 Singapore Telecom (d,j) Singapore 773 643 20.2% 480 223 22 Telia AB (i) Sweden 702 697 0.7% 683 831 23 DGT Taiwan (a) Taiwan 593 498 19.1% 441 242 24 Worldcom (b,m) United States 547 278 95.7% n.a. n.a. 25 TeleDanmark Denmark 533 488 9.2% 452 362 26 Etisalat U.A.E. 504 428 17.8% 342 242 27 Saudi Com. Ministry Saudi Arabia 499 477 4.6% 455 320 28 OTE (a) Greece 468 423 10.6% 336 213 29 Norwegian Telecom Norway 432 396 9.1% 376 281 30 Telekom Malaysia Malaysia 408 342 19.3% 258 140 31 Telecom Eireann (a,d,I) Ireland 407 324 25.6% 316 262 32 Korea Telecom Rep. of Korea 404 327 23.5% 285 188 33 Telekomunikacja Polska Poland 381 357 6.7% 273 81 34 Turkish P’R Turkey 374 284 31.7% 265 159 35 Videsh Sanchar (d,k) India 341 314 8.6% 284 147 36 Telebras Brazil 319 199 60.5% 182.4 165 37 Rostelcom Russia 287 229 25.4% 201 n.a. 38 IDC (d) Japan 282 283 7.2% 239 56 39 Telkom South Africa South Africa 280 263 6.5% 255 156 40 ITJ (d) Japan 270 251 7.6% 228 61 Mi-l-r is Minutes of Telecommunications Traffic. Data are for public voice circuits only rounded to the nearest mdlion MITE. a. Data from 1993 forward based on billing pmnt of call, not originat- ing point. b. Data for North American carriers include cross-border traffic. c. For Deutsche Telekom, all data ~nclude outgoing traffic from the former East Germany. d. Data are for the Fiscal Year (April 1995 to March 1996). HKT and Mercury are majority owned by Cable & W~retess (U.K.). e. Combined totals for Iritel and Italcable. Prior to 1994, Iritel (for- merly ASST) handled intra-cont~nental traffic only, and Italcable carried overseas traffic. f. Includes Hong Kong-China traffic. g. Stentor was formerly Telecom Canada; Stentor traffic is for U.S. only of which approximately 70% is originated by Belt Canada. h. Telstra was formerlyAOTC. i. Telia AB was formerlyTeleverket. j. Singapore Telecom data, exceptfor 1990, include traffic to Malaysia (except local border traffic). k. Videsh Sanchar data exclude traff=c to Bangladesh, Nepal, Pakistan and Sri Lanka. t. Telecom E~reann data exclude traffic to Northern Ireland. m. 1995 WorldCom data reflects full year data from tDB, LDDS and WdTel acquisitions. © Telel3eography, Inc. 1996 © TeleGeography, Inc. 1996 Top International Routes and Carriers Table 4. The Top 50 International Routes, 1995 Countries Mi’n" each way Total Mi’n" 1. United States/Canada 2998.0/2063.7 5061.7 2. United States/Mexico 1915.3/833.9 2749.2 3. United States/United Kingdom 1017.4/678.1 1695.5 4. Hong Kong/China 913.6/750,0 1663.6 5. United States/Germany 657.7/290.3 948.0 6. United States/Japan 574.3/319.1 893,4 7. Switzerland/Germany 408.2/383.0 791.2 8. Germany/Austria 416.3/368.6 784.9 9. Germany/United Kingdom 365.1/364.4 729.5 10. Germany/France 389.8/325.9 715.7 11. United Kingdom/France 360,8/317.7 678.5 12. Netherlands/Germany 339.5/335.9 675.4 13. Germany/Italy 375.5/299.0 674.5 14. United Kingdom/Ireland 371.4/264.0 635.4 15. United States/France 355.4/180.9 536.3 16. Germany/]’urkey 375.6/124.5 500.1 17. Belgium/France 263,0/228.2 491.2 18. France/Italy 242.1/239.0 481.1 19. Netherlands/Belgium 238.6/229.8 468.5 20. United States/Korea 312.3/141.0 453.0 21. Switzerland/France 280.0/154,6 434.6 22. Switzerland/Italy 252.8/181.8 434.6 23. United States/Dominican Republic 342.9/86.3 429.2 24. Germany/Poland 278.9/146.4 425.3 25. United States/Hong Kong 314.1/102.7 416,8 26. Singapore/Malaysia 218.0/184.0 402.0 27. United States/Taiwa n 273.2/111.0 384.2 28. United States/Brazil 277.6/101.3 378.9 29. United States/Italy 273.4/103.6 377.0 30. United States/Philippines 294.8/62.0 356.8 31. Netherlands/United Kingdom 179.9/173.5 353.4 32. United States/Australia 200.1/144.0 344.1 33. France/Spain 179.7/158.2 337.9 34. United Kingdom/Italy 188.0/149.3 337.3 35. United States/India 284.1/49.9 334.0 36. Germany/Spain 170.4/160.7 331.1 37. United Kingdom/Spain 171.1/147.0 318.1 38. United States/Colombia 253.2/58.8 312.0 39. Australia/United Kingdom 182.0/127.3 309.3 40. Australia/New Zealand 142.0/142.0 284.0 41. United States/Israel 213.4/70.3 283.7 42. United States/China 230,2/52.3 282.5 43. Germany/Belgium 144.4/137.4 281.8 44. Japan/Korea 150.3/126.0 276.3 45. Canada/United Kingdom 142.0/120.5 262.5 46. Japan/China 171.0/88.8 259.8 47. Taiwan/China 140.9/115.0 255.9 48. United States/Netherlands 161.3/89.7 251.0 49. Norway/Sweden 111.0/124.0 235.0 50. Sweden/Finland 128.0/99.0 227.0 All data in millions of minutes of telecommunications traffic (MITT). The country which generates more traffic on each route is listed first. The routes listed above total 31.7 billion minutes, 53 percent of all international traffic. For routes to and from the United States, calls are measured by point of billing in both directions (see Methodology). TeleGeo{~raphy 1996/97 © TeleGeography, Inc. 1996 Table 5. Country/Carrier United States AT&T Sprint Worldcom Others Market Share of Competing International Carriers, 1988-1995 Percentage of Outgoing MiTT 1988 1989 1990 1991 1992 1993 1994 1995 89.1 83.3 78.4 74.8 70.3 62.2 60.1 54.3 7.0 10.2 14.6 17.8 21.2 24.8 26.5 28.5 3.5 5.8 6.4 6.3 7.3 10.3 11.1 11.3 n.a. 0.6 2.1 3.5 0.2 2.4 United Kingdom* BT 95.5 91.0 86.0 81.0 76.8 74.2 68.6 67.7 Mercury 4.5 9.0 t4.0 19.0 23.2 24.0 28.1 25.8 IPL Resellers 2.2 3.3 6.5 Japan* KDD 93.3 88.0 73.3 69.7 66.9 66.3 66.2 IDC 3.7 6.5 13.3 15.3 16.9 17.3 17.3 ITJ 3.0 5.5 13.4 15.0 16.2 16.4 16.5 New Zealand* TNZ 92.0 82.0 80.0 78.4 74.8 78.0 ClearCorn 8.0 18.0 20.0 21.6 25.2 22.0 Korea, Republic of Korea Tetecorn 79.9 74.5 68 7 72.6 Dacom 20.1 25.5 31.3 27.4 Chile Entel Chile 80.0 55.0 36.3 36.5 Chitesat 20.0 20.0 24.8 23.1 VTR Telecom 1.0 5.0 24.2 7.4 CTC-Mundo 12.8 20.2 BellSouth Chile 1.5 9.9 tusatel 0.1 1.7 CNT 0.3 0.5 Philippines PLOT 91.6 84.2 69 68 Philippine Global Corn 8.4 15.8 23 23 Eastern Telecom n.a. n.a. 7 6 Capitol Wireless n.a. n.a. 1 1 Smar[ 1 ICC 1 Notes: MzFF is Minutes of Telecommunications Traffic. Data based on outgmng znternational traffic for the public switched network only. Unless stated, data exclude traffic and market share of carriers reselhng internat=onal private line serv=ces (IPL resellers). Market shares are for the full year, begznning =n the first year of competit=on. In 1995, some U,S. facilities-based international carners also resold the international switched services of other U.S. famht~es-based carners as fol- lows: MCl, 50.5 million minutes; Sprint, 39.6 milhon minutes; WorldCom, 423.3 million minutes. Market shares for U.S. carriers prior to 1994 exclude resellers and, prior to 1993, traffic to Canada and Mexico; for the traff=c base of second tzer U S. carriers, see page 87. U.K. carriers’ traffic to Ireland is excluded prior to 1994. In 1993, Chilean shares do not add up to 100% because Chilesat reportedly acted as an international gateway in 1993. The 1994 and 1995 market shares for Chzle are based on traffic for the month of December only. Market shares based on fiscal year reporting. ©TeleGeography, Inc. 1996 © TeleCeography, Inc. 1996 Market Share of Competing International Carriers Table 5. Country/Carrier Australia Telstra Optus IPL Rese!lers Market Share of Competing International Carriers, 1988-1995 (continued) Percentage of Outgoing MiTT 1988 1989 1990 1991 1992 1993 1994 1995 98.0 87.0 76.3 73.4 2.0 13.0 21.9 23.4 1.8 3.2 Canada (Canada-U.S. route only) Stentor 93 80 65 Unitel 2 8 8 Weste! 1 1 1 IPL Resellers 4 11 26 Dominican Republic Codetel >90 85.8 83.0 Tricom n.a 6.7 7.5 All America Cables and Radio, Inc. (AACR) n.a. 7.5 9.5 Sweden Telia AB 92.3 86.9 78.0 Tete-2 7.7 13.1 22.0 Finland Tetecom Finland 90 72.8 Finnet International 5 19.1 Telivo 3 7.7 Others 2 0.4 Indonesia PT Indosat 99 >95 PT Satelindo 1 5 Notes: M~TT zs Minutes of Telecommumcatmns Traffic. Data based on outgoing internatzonal traffic for the pubhc switched network only Unless stated, data exclude traffic and market share of comers reselhng mternatzonal private line servzces (IPL resellers) Market shares are for the full year, begmmng m the first year of competition. For Austraha, market shares for 1994 and 1995 are baaed on traffzc for October to December quarters only and reflect wholesale minutes for fac~l~- tzes-based carriers only. For Canada, market shares reflect an estzmated 200 milhon minutes of traffic handled by IPL resellers which ~s not reflected m offzczal car- nor reports. For Indonesza, PT Satelindo only began internatzonal servzce m September 1994; the 1995 figures reflect June 1995 market estimates ©TeleBeography, Inc. 1996 TeleGeography1996/97 © TeleGeographyo Inc. 1996 International Traffic Statistics TeleGeography 1996/97 © TeleGeography, Inc. 1996 Americas International Traffic TeleGeography 1996/97 © TeleGeography, Inc. 1996 © TeleGeography, Inc. 1996 International Telephone Traffic--Americas Argentina Destination Largest Telecommunications Routes, Mi3-r 1. United States .......... 35.4 2. Uruguay .............. 32.7 3. Brazil ................ 21.9 4. Chile .................. 12.6 5. Spain ................ 11.4 6. Italy ................... 9.3 7. Paraguay .............. 7.8 8. Peru ................... 5.3 9. Bolivia ................ 4.8 10. France ................ 4.0 1995 Percentage of Outgoing Traffic i~i~!~ 19.7% 7.0% 2.2% 11. Germany .............. 3.5 12. Mexico ................ 3.5 ~!!?,,~: 1.9% 13. United Kingdom ......... 3.4 ~’1.9% 14. Israel ................. 2.9 ~i~ii,~.’. 1.6% 15. Colombia .............. 2.2 i:i~, 1.2% 16. Canada ................ 2.1 i?:i:i 1.2% 17. Venezuela ............. 1.9 ~i:~,’ 1.t% 18. Ecuador ................ 0.9 o.5% 19. Netherlands ............ 0.8 ~; o.5% 20. Japan ................. 0.7 ~, o.4% Other .12.4 ~;~,~ ~,;~ ~:~:~:~ 6.9% Mi~ is Minutes of Telecommunications Traffic. Data are in millions of minutes for public voice circuits. TeleGeography, Inc. 1996 National Traffic Balance MiTT 1993 1994 1995 Incoming 192.3 252.6 299.4 Outgoing 137.1 175.0 179.4 Surplus (Deficit) 55.4 77.7 119.9 Total Volume 39.4 427.6 478.8 Note: Data based on billing point of traffic. TeleGeography 1996/97 © TeleGeography, Inc. 1996 Destination Largest Telecommunications Routes, 1. United States ........... 4.8 2. Argentina .............. 3.3 3. Brazil .................. 3.2 4. Chile .................. 2.3 5. Peru ................... 1.6 Percentage of Outgoing Traffic 1995 6. Germany .0.5 ~,,:~,’>,~>,~ 2.4% ............. 7. Spain ................. 0.4 ~;1.9% 8. Colombia .0.4 ~,, ~,~,~,~,~ 1.9% .............. 9. Mexico .0.4 ~;~i 1.9% 10. Paraguay .0.3 ~i 1.4% .............. .~ 11. Italy .0.3 ................ .. ~: 1.4% 12. Venezuela .0.2 ~ii 1.o% ............. 13. United Kingdom ......... 0.2 ~! 1.0% 14. France ................ 0.2 ~!! 1.o% 15. Japan .0.2 ;~’ 1.0% 16. Canada ................ 0.2 Ii~ii; 1.o% 17. Ecuador .0.2 ...... 1.0% ............... 18. Switzerland ............. 0.2 ~i 1.o% ~9. Uruguay ............... 0.2 I;~!~:,~; 1.o% 20. Panama .0.2 1.o% Other ................. 1,5 MiTF is Minutes of Telecommunications Traffic. Data are in millions of minutes for public voice circuits. © TeleGeography, Inc. 1996 National Traffic Balance MiTT 1993 1994 1995 Incoming n.a. n.a. 49.2 Outgoing 16.6 18.0 20.8 Surplus (Deficit) n.a. n.a. 28.4 Total Volume n.a. n.a. 70.0 Nete: Data based on billing point of traffic. © TeleGeography, Inc. 1996 International Telephone Traffic--Americas Brazil Destination 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. Largest Telecommunications Routes, 1995 Mi’l’]" Percentage of Outgoing Traffic ~;~:~’~.~:~, ~: -~ :/~’ 7.6% United States ......... 100.2 Argentina ............. 24.4 Italy .................. 14.2 Germany .............. 13.8 ~;i~’~" 4.3% Portugal .............. 13.4 United Kingdom ....... .11.5 ~i~;~:~:; 3.6% France .10.3 ~~’;~ 3.2% Uruguay .7.8 ~i,’;%:~ 2.5% .............. Japan ................. 7.7 !i~i~ 2.4% Spain ................. 7.1 ~!i, 2.2% Lebanon ............... 6.8 ~" 2.10/o Paraguay .............. 6.4 2.0’/o Chile .................. 6.1 ~!i’~:i" 1.90/o Moldova ............... 5.4 i’~!i!ii~’ 1.7% Switzerland ............ 4.7 i~,~i. 1.5% Canada ................ 4.6 !i~i,’" 1 4% Sao Tome and Principe ...4.6 il;~:’ 1.4°/o Bolivia ................ 4.6 ~il 1.4% Israel .................. 4.6 "iiiii’: 1.40/o Mexico ................ 4.0 ;~; 1.2% MiTT is Minutes of Telecommunications Traffic. Data are in millions of minutes for public voice circuits. © TeleGeography, Inc. 1996 National Traffic Balance MiTT 1993 1994 1995 Incoming 373.8 408.0 495.5 Outgoing 182.4 199.0 319.4 Surplus (D eficit) 191.4 209.0 176.1 Total Volume 556.2 607.0 814.8 Note: Data based on billing point of traffic. TeleGeography 1996/97 © TeleGeography, Inc. 1996 Largest Telecommunications Routes, 1995 Destination MiTT Percentage of Outgoing Traffic 1. United States ........ 2061 ~~ ~i~;;~ 69.7% 2. United Kingdom ....... 142 4.8% 3. France ............... 47 ~il 1.6% 4. Germany .............. 41 ~ii 1.4% 5. Hong Kong ............ 40 ~i 1.4% 6. India ................. 36 ~.2% 7. Italy .................. 35 i;.il ~.2% 8. Philippines ............ 30 ~.0% 9. Australia .............. 29 ~i ~.0% 10. Jamaica .............. 26 ~i 0.9% 11. Japan ................ 21 o.7% 12. Netherlands ........... 18 i 0.6% 13. Mexico ............... 18 i 0.6% 14. Poland ................ 16 0.5% 15. Switzerland ............ 15 "o.5% Other ................ 384 ~i’.~!.~i~i~:~’~:~i’i:i!~ 13.0% MiTr is Minutes of Telecommunications Traffic. Data are in millions of minutes for public voice circuits. © TeleGeography, Inc. 1996 National Traffic Balance MiTT 1993 1994 1995 Incoming 503.4 543.8 603.4 Outgoing 761.5 861.2 897.9 Surplus (Deficit) (258.1) (317.4) (294.5) Total Volume 1264.9 1405.0 1501.3 Notes: Incoming and outgoing totals are for T~l~globe only and exclude all Canada-U.S. traffic. T~l~globe data based on billing point of traffic. U.S. route traffic is for Stentor, Unitel and IPL rese!lers combined, but IPL resellers’ traffic is not included on other routes (i.e., to the U.K. and Australia). For further details, see Methodology Section. Route data are rounded to the nearest million minutes. © TeleGeography, Inc. 1996 International Telephone Traffic--Americas Chile Largest Telecommunications Routes, 1995 Destination Mi’l-r Percentage of Outgoing Traffic 1. United States .......... 45 2. Argentina .23 ~’~’~"~ ’-~ ,’~-~-~’~:~i~~’~, 16.9% 3. Brazil ................ 11 4. Spain ................ 10 ~:;~::’;’~;;’:~;}~;~:~,._,.. ~,~ 7.4% 5. Peru ................... 6 ~,~:,,~ 4.4% 6. Canada ................ 4 ~:/2.9% 7. Germany .4 ~:’ ~:~;: ............. ~,,~ % 2.9% 8. Mexico .4 ~:;;:~;~’~: ............... C~,,~’ 2.9% 9. Bolivia ................. 3 ~;"~ ,~ ,;’~ 2.2% 10. France ................. 3 ~::~ 2.2% 11. United Kingdom ......... 3 ~z~:, 2.2% 12. Ecuador ................ 3 ~::~:’ 2.2% t3. Italy ................... 3 @: 2.2% Other ................. 14 ;’~-,_ "~:~,~_~;~ 10.3% MiTT is Minutes of Telecommunications Traffic. Data are in millions of minutes for public voice circuits. © TeleGeography, Inc. 1996 National Traffic Balance MiTT 1993 1994 1995 Incoming 105.0 n.a. n.a. Outgoing 61.7 73.5 136.9 Surplus (Deficit) 43.3 n.a. n.a. Total Volume 166.7 n.a. n.a. Nnte: Bata rounded to the nearest million minutes. Incoming traffic for Entel, VTR and CTC totaled 134.1 million minutes in 1995. Other carriers terminated approximately 30-40 million minutes. TeleGeography 1996/97 © TeleGeography, Inc. 1996 Largest Telecommunications Routes, Destination MiTr 1. United States .......... 58.8 2. Venezuela ............ 14.2 3. Spain ................. 5.4 4. Ecuador ............... 5.3 5. Panama ............... 4.6 3.7% 6. Mexico ................ 4.6 @ 3.6% 7, Italy ................... 3.7 ~i 2.9% 8. United Kingdom ......... 3.0 2.4% 9. Costa Rica ............. 2.6 2,0% 10. France ................ 2,5 ~!ii 2.0% 11. Germany .............. 2.5 ~i2.0% 12. Brazil ................. 2.3 ~! 1.8% 13. Peru ................... 2.3 ~!: 1.8% 14. Argentina .............. 2.2 ~!~i~ 1.7% 15. Canada ................ 1,9 !~i!i! 1.5% 0t h e r ................. 11,2 ~’:~;;~~~:;:: ~!~,~ ~:~,~,, .~.;..~-. 8.8% 1995 Percentage of Outgoing Traffic MiTT is Minutes of Telecommunications Traffic. Data are in millions of minutes for public voice circuits. © TeleGeography, Inc. 1996 National Traffic Balance MiTT 1993 1994 1995 Incoming 278.7 302.8 351.5 Outgoing 102.4 120.3 127.3 Surplus (Deficit) 176.3 182.5 224.2 Total Volume 381.1 423.1 478.8 Note: Totals may appear inconsistent with other figures due to rounding. Data based on billing point of traffic. © TeleGeography, Inc. 1996 International Telephone Traffic--Americas Dominican Republic Largest Telecommunications Routes, 1995 Destination MiTT 1. United States .......... 60.2 2. Puerto Rico ............. 9.9 3. Spain ................. 1.8 4. Germany .............. 1.5 5. Italy ................... 1.4 6. Canada ................ 1.2 7. Venezuela ............. 0.8 8. Mexico ................ 0.6 9. Colombia ............... 0.5 10. Cuba .................. 0.5 11. Switzerland ............ 0.5 12. Panama ............... 0.4 13. Haiti ................... 0.4 14. Argentina .............. 0.3 15. France ................ 0.3 16. Curacao ............... 0.3 17. United Kingdom ......... 0.3 18. Costa Rica ............. 0.3 19. Netherlands ............ 0.2 20. Austria ................. 0.2 Other ................. 3.7 Percentage of Outgoing Traffic ~:,, ,,~:~;, 11.6% ,2.1% if 1.8% }i~ 1.6% i" 1.4% ii 0.9% O.7% 0,6% 0.6% O.6% 0.5% O.5% 0.4% 0.4% 0.3% 0,3% 0.3% 0,3% 0.2% MiTT is Minutes of Telecommunications Traffic. Data are in millions of minutes for public voice circuits. © TeleGeography, Inc. 1996 National Traffic Balance MiTT 1993 1994 1995 Incoming n.a. 404.0 424.1 Outgoing 58.3 63.5 85.4 Surplus (Deficit) n.a. 340.5 338.7 Total Volume n.a. 467.5 509.4 Note: Data are for Codetel only and are based on billing point of traffic. In 1995, AACR had 9.4 million minutes of outbound traffic and 59.1 million minutes inbound. TeleGeography 1996/97 © TeleGeography, Inc. 1996 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. Largest Telecommunications Routes, 1995 Destination MiTr Percentage of Outgoing Traffic United States .19.1 :’,G,7 ,~ ;g~:~,,,,~?,~;~~!~,~ 51.3% Colombia .5.0 ~7~:~: ........ 13.4% Chile .................. 1.3 ,~Z 3.4% Venezuela ............. 1.0 ~7 2.8% Peru ................... 0.9 :~; 2.4% Brazil ................. 0.9 ~1~ 2.4% Spain ................. 0.9 ~:~ 2.4% Mexico ................ 0.7 ’ 1.9% Argentina .............. 0.7 :’7 1.9% Panama ............... 0.7 1.8% Italy ................... 0.6 1.7% Canada ................ 0,6 1.6% Germany .............. 0.6 1.5% France ................ 0.4 1.o% United Kingdom ......... 0.4 1.0% Costa Rica ............. 0.3 0.8% Swi~erland ............ 0.3 o3% Bolivia ................ 0.2 o.5% Japan ................. 0.2 o.4% Hong Kong ............. 0.2 o4% Other ................. 2.4 Mi~ is Minutes of Telecommunications Traffic. Data are in millions of minutes for public voice circuits. © TeleGeography, Inc. 1996 National Traffic Balance MiTT 1993 1994 1995 Incoming 102.3 128.6 154.8 Outgoing 33.6 36.4 37.2 Surplus (Deficit) 68.7 92.2 117.6 Total Volume 136.0 165.0 192.0 Nete: Data based on billing point of traffic. Totals may appear inconsistent with other figures due to rounding. © TeleGeography, Inc. 1996 International Telephone Traffic--Americas Mexico Destination Largest Telecommunications Routes, 1995 Mi~’l" Percentage of Outgoing Traffic 1.5% 1.2% 1.0% 0.6% 0.6% 0.6% 0.6% 0.5% 0.5% 0.5% 0.4% 0.3% 0.3% O.3% 0.3% 0.3% 0.2% 0,2% 0.2% ~’" 2.3% MiTT is Minutes of Telecommunications Traffic. Data are in millions of minutes for public voice circuits. © TeleGeography, Inc. 1996 National Traffic Balance 1. United States ......... 831.8 2. Canada ............... 14,2 3. Spain ................ 11.4 4. France ................ 9.4 5. Cuba .................. 6.1 6. Germany .............. 5.9 7. Colombia .............. 5.8 8. United Kingdom ......... 5.4 9, Argentina .............. 4,8 10. Guatemala ............. 4.6 11. Italy ................... 4.4 12. Brazil ................. 3.8 13. Costa Rica ............. 3.1 14. Chile .................. 3.0 15. Venezuela ............. 3.0 16. Japan ................. 2,4 17. Peru .................. 2.4 18. El Salvador ............. 2.2 19. Israel .................. 2.1 20. Switzerland ............ 1.9 Other ................. 22.3 MiTT 1993 1994 1995 Incoming 1,370.6 1,829.4 2,114.0 Outgoing 625.4 844.1 950.0 Surplus (Deficit) 745.2 985.4 1,164.0 Total Volume 1,996.0 2,673.5 3,064.0 Note: Data based on billing point of traffic. 1993 figures do not include traffic generated by Tel~fonos del Noroeste (Telnor), a Telmex subsidiary. Totals may appear inconsistent with other figures due to rounding. TeleGeography 1996/97 © TeleGeography. Inc. 1996 Largest Telecommunications Routes, 1995 Destination MiTT 1. Argentina .............. 6.4 2. Brazil ................. 4.4 3. United States ........... 2.8 4. Chile .................. 1.1 5. Uruguay ............... 1.0 Percentage of Outgoing Traffic 6. Italy ................... 0.7 7. Germany .............. 0.4 8. Spain ................. 0.4 9. Taiwan ................ 0.3 10. Bolivia ................ 0.3 11. Korea, Rap. of .......... 0.3 12. Peru ................... 0.2 13. France ................ 0.2 ,, 1.o% 14. Japan ................. 0.2 :." 0.8% 15. Hong Kong ............. 0.2 ’~~ o.8% 16. United Kingdom ......... 0.1 0.6% 17. Switzerland ............ 0.1 o.6% 18. Canada ................ 0.1 0.6% 19. Mexico ................ 0.1 20. Panama ............... 0.1 o.5% Other ................. 0.2 ii’:, 1.o% ~: :;~): 5.2% ~,:’ 2.1% i ’:’~ :~ 1.8% 1.3% 1.2% MiTT is Minutes of Telecommunications Traffic. Data are in millions of minutes for public voice circuits. © Tele6eography, Inc. 1996 National Traffic Balance MiTT 1993 1994 1995 Incoming 24.5 30.6 n.a. Outgoing 15.5 18.1 19.4 Surplus (Deficit) 9.0 12.5 n.a. Total Volume 40.0 48.7 n.a. Nate: Totals may appear inconsistent with other figures due to rounding. © TeleGeography, Inc. 1996 International Telephone Traffic--Americas Peru Largest Telecommunications Routes, 1995 Destination MiTI" 1. United States .......... 25.5 2. Chile .................. 5.5 3. Argentina .............. 3.5 4. Spain ................. 3.5 5. Italy ................... 2.0 Percentage of Outgoing Traffic #%~’;,,~ ",~ 5.2% ~,’~ 3.0% 6. Brazil ................. 2.0 7. Colombia .............. 2.0 8. Venezuela .1.5 "~,~’ 2 ° ............ .2’/o 9. Bolivia ................ 1.5 10. Mexico ................ 1.5 !~,~:~ 2.2~/o 11. Ecuador ............... 1.5 12. Japan ................. 1.5 fi!~:~i: 2.2% 13. Germany .............. 1.0 !i~il.5% 14. Canada ................ 1.0 iii,,i 1.5°/o 15. United Kingdom ......... 1.0 ~:;,!~, 1.5% MiTT is Minutes of Telecommunications Traffic. Data are in millions of minutes for public voice circuits. © Tele6eography, Inc. 1996 National Traffic Balance MiTT 1993 1994 1995 Incoming 152.4 178.6 195.4 Outgoing 39.0 51.0 66.7 Surplus (Deficit) 113.4 127.6 128.7 Total Volume 191.4 229.6 262.1 Note: Data based on billing point of traffic and rounded to the nearest 0.5 million. Totals may appear inconsistent with other figures due to rounding. TeleGeography 1996/97 © TeleGeography, Inc. 1996 United States Ou oin Destination Largest Telecommunications Routes, 1995 MiTr 1994 MiTI" 1995 Percentage of Outgoing Traffic 1995 1. Canada ............ 2,635.2 .... 2,998.0 2. Mexico .1,654.3 .1,915.3 .............. ~;’~,~L~ ~,~ 12.3% 3. United Kingdom .905.5 .1,017.4 F~:’~’~?~,~,-~,-~’~"~ 6.5% ... .. 4. Bermany ............ 603.3 ...... 657.7 5. Japan ............... 465.6 ...... 574.3 6. France .304.5. .355.4 a~a~t~k o .......... ...... 7. Dominican ~epublic ...... .309.7 .342.9 ~,~a~ 2.2% 8. Hon9 Kon9 .213.3 ...... 314.1 .......... 9. Korea, ~ep. of ........ 282.7 ...... 312.3 .... 10. Philippines ........... 258.6. .294.8 11. India ................ 188.6 ...... 284.1 ~"~}} ~.8% 12. Brazil .221.5 .277.6 ~’~ ~.e% 13. Italy .250.4 .273.4 ............ .. .... ~;~L: ~.7% 14. Taiwan .............. 225.6 ...... 273.2 15. Colombia ............ 229.2 ...... 253.2 ~F{~1.6~ 16. China .169.2. .230.2 17. Israel ............... 195.4 ...... 213.4 1.4% 18. Australia ............ t54.4 ...... 200.1 19. Jamaica ............. 167.3 ...... 186.3 20. Netherlands .......... 110.5 ...... 161.1 ~:~ 1.0% Other .3,636.0 .4,488.2 MiTT is Minutes of Telecommunications Traffic. Data are in millions of minutes for public voice circuits. © TeleGeography, Inc. 1996 National Traffic Balance MiTT 1993 1994 1995 Incoming 5,342.8 6,133.1 7,010.6 Outgoing 11,392.2 13,200.3 15,623.0 Surplus (Deficit) (6,049.4) (7,067.2) (8,612.4) Total Volume 16,735.0 19,333.4 22,633.6 Note: Data based on billing point of traffic. Totals may appear inconsistent with other figures due to rounding. Totals for 1995 include IPL reseller traffic. Data excludes non-Continental U.S. territories (Puerto Rico, Virgin Islands and Guam). © TeleGeographyo Inc. 1996 International Telephone Traffic--Americas United States (Incoming) Largest Destination Mi’l-r 1994 MiTT 1995 1. Canada ............ 1,688.1 .... 2,063.7 2. Mexico .............. 747.0 ...... 833.9 3. United Kin9dom ...... 588.7 ...... 678.1 4. ,Japan ............... 304.7 ...... 319.1 5. Germany ............ 275.8 ...... 290.3 6. France .............. 170.0 ...... t80.9 7. Australia ............ 138.4 ...... 144.0 8, Korea, Rep, of ........ 123.5 ...... 140,7 9. Taiwan ............... 93.4 ...... 108.4 10. Italy ................ 101.0 ...... 103.6 11. Hong Kong ........... 100.5 ...... 102.7 12. Brazil ................ 61.8 ...... 101.3 13. Netherlands .......... 82.3 ....... 89.7 14. Dominican Republic ...60.5 ....... 86.3 15. Switzerland ........... 72.8 ....... 77.0 16. Israel ................ 59.8 ....... 70.3 17. Sweden .............. 58.2 ....... 62.0 18. Colombia ............. 58.1 ....... 58.8 19. Venezuela ............ 54.9 ....... 56.0 20. China ................ 48.4 ....... 52.3 Other ........................ 1,391.5 MiTt is Minutes of Telecommunications Telecommunications Routes, 1995 Percentage of Incoming Traffic 1995 4.8% 2.1% ~: 1.5% 1.5% 1.5% ~.4% 1.3% ~t~, 1.2% 0.9% 0.8% ~:~ 0.7% Tragic. Da~a are in millions of minu~es for public voice ~ircui~s. TeleGeography, Inc. 1996 Note: Data based on billing point of traffic. Incoming traffic reported by the United States may net match outgo- ing traffic to the United States reported by other countries due to different accounting procedures (some coun- tries may report U.S.-billed calls to the U.S. as outgoing calls to the U.S.), different fiscal years, and inclusion or exclusion of operator-assisted calls. See Methodology, at the end of this section, for more information. TeleGeography 1996/97 © TeleGeography, Inc. 1996 USA: Other Correspondents © TeleGeography, Inc. 1996 International Telephone Traffic--Americas USA: Traffic by Carrier Market Share of International Traffic by Route, 1995 U.S. Billed Traffic AT&T MCI Sprint WorldCom Canada 54.51 25.82 12.39 2.26 Mexico 62.58 23.91 9.83 3.65 U.K. 54.65 26.34 11.73 5.31 Germany 58.15 26.37 12.18 2.82 Japan 48.45 33.62 11.88 4.75 Philippines 62.91 28.26 7.27 1.31 France 48.57 31,05 12.91 7.17 Italy 61.08 22.61 12.70 3.55 Korea, Rep. of 52.99 31.50 11.79 3.61 Brazil 56.20 24.47 8.80 10.50 India 48.97 41.72 7.59 0.31 Dominican Republic 39.10 42.40 4.66 7.74 Foreign Billed AT&T MCI Sprint WorldCom 54,97 23.99 7.40 1.91 63.22 25.42 8.62 2.72 53.90 26.28 12.09 6.61 55.58 27,79 12.87 3.25 53.34 27.79 12.12 5.32 56.68 31.35 8.28 0.77 49.89 26.57 14.23 9.13 60.25 23.44 12.88 3.18 54.06 31.05 10.07 3.20 62.31 22.46 9.33 5.88 45.48 46.46 6.55 0.00 87.75 20.56 5.27 6.41 57.36 30.45 6.06 6.09 76.41 16.99 6.59 0.00 40.68 40.63 13.71 3.59 48.35 36.08 12.04 2.90 52.74 25.59 15.21 6.44 38.30 36.25 15.61 8.08 54.22 23.73 14.57 6.85 52.73 30.57 11.74 4.93 56.09 22.97 13.34 7.45 56.46 28.47 10.83 4.05 48.97 27.47 13.58 9.43 57.52 26.04 10.02 6.37 58.52 25 94 12.15 3.39 Colombia 52.32 34.49 6.85 6.31 Jama=ca 61.94 31.17 7.47 0.00 Taiwan 41.60 39.97 13.09 4.45 China 48.33 35.40 12.01 4.18 Israel 50.65 30.29 14.96 4.09 Hong Kong 30.28 31.95 33.44 4.32 Australia 47.25 22.00 17.65 6.97 Argentina 52.54 29.18 10.61 7.66 Netherlands 47.83 23.36 14.59 6.98 Spain 56.45 22.59 11.80 7.34 Switzerland 51.28 26.61 14.79 7.24 Venezuela 56.14 26.89 10.70 6.25 Sweden 51.90 28.97 13.50 4.80 © TeleGeography, Inc. 1996 Traffic Carried by Second Tier U.S. Facilities-Based International Carriers, 1995 Outbound Inbound Top Outbound Resold Out- Carrier Minutes (m) Minutes (m) Routes (Minutes) bound (Minutes) !80.0 229.9 0 72.4 -- 0 38.6 1.4 89.7 fONOROLA Canada (180) International Telecom Corp. (ITC) Dora. Rep. (18.5) Pacific Gateway Exchange (PI3E) Australia (12.1) Netherlands (11.5) Japan (7.0) WorldXchange (formerly CTS) 24.2 4.8 Canada (7.9) 114.6 U.K. (6.2) 6TE Hawaiian Telephone Co. 19.2 28.3 Philippines (4.8) 0 Japan (4.9) Cable & Wireless, Inc. (CWl) 10.9 -- Italy (2,4) 291.2 U.K. (5.1) ACC Global 6.9 -- U.K, (6.9) 0 MFS International 6.8 -- U.K. (5.1) 0 Startec, Inc. 4.1 0.7 India (4.1) 6.7 Esprit U.K. 2.9 -- U.K. (1.4) 0 Total 366.0 265.1 Note: All data ~n milbons of m~nutes based on bflhng point of call. Carriers and traffic from off-shore U.S points (=.e., Puerto Rico, Virgin Islands, Guam) are excluded. Data includes traffic carried on International S=mple Resale (ISR) facilities. The last column of the table refers to outbound minutes b~lled on a resale bas=s (~.e ,the traff=c was actually transported over international famht=es by a third path/such as AT&T) These rn~nutes are thus included =n the traff=c base of other U S. careers m Table 3 on p. 64 and in Table 5 on pp. 66-67 © TeleGeography, Inc. 1996 TeleGeography 1996/97 © TeleGeography, Inc. 1996 Destination Largest Telecommunications Routes, 1. Argentina ............. 27.7 2. Brazil ................. 6.8 3. United States ........... 5.4 4. Spain ................. 1.9 5. Chile .................. 1.3 1995 Percentage of Outgoing Traffic ~#~;~{~’~ 13.5% 6. Paraguay .............. 1.0 7. Italy ................... 0.8 8. France ................ 0.5 9. Germany .............. 0.5 10. United Kingdom ......... 0.4 11. Canada ................ 0.4 12. Mexico ................ 0.4 13. Israel ................. 0.3 14. Venezuela ............. 0.3 15. Switzerland ............ 0.3 16. Australia .............. 0.2 17. Colombia .............. 0.2 18. Peru .................. 0.2 19. Panama ............... 0.2 20. Sweden ............... 0.1 Other ................. 1.2 3.7% ~i 2.5% ~iii 2.0% 1.6% 1.1% 0.9% il 0.7% 0.7% !0.7% i 0.7% i~ 0.6% i 0.6% 0.5% 0.4% 0.4% 0.3% 0.3% #:~i 2"4% MiTT is Minutes of Telecommunications Traffic. Data are in millions of minutes for public voice circuits. © TeleGeography, Inc. 1996 National Traffic Balance MiTT 1993 1994 1995 Incoming 58.0 67.7 73.9 Outgoing 37.4 46.3 49.9 Surplus (Deficit) 20.6 21.4 24.0 Total Volume 95.4 114.0 123.8 Note: Data based on billing point of traffic. Totals may appear inconsistent with other figures due to rounding. © TeleGeagraphy, Inc. 1996 International Telephone Traffic--Americas Venezuela Destination Largest Telecommunications Routes, 1995 MiTT Percentage of Outgoing Traffic ..-":~,. "~. ~,--: "~:~’~ ~,,~- .. ’. ; ...... 42.9% 1. United States .......... 55.3 2. Colombia ............. 17.9 3. Spain ................. 7.9 4. Italy ................... 5.9 5. Peru ................... 3.1 6. Canada ................ 2.7 7. Portugal ............... 2.5 8. Brazil ................. 2.5 9. Mexico ................ 2.3 10. Argentina .............. 2.2 11. Dominican Republic ..... 2.1 12. France ................ 2.1 13. Ecuador ............... 2.0 14. Germany .............. 1.7 15. United Kingdom ......... 1.7 16. Chile .................. 1.6 17. Puerto Rico ............. 1.4 18. Netherlands Antilles ..... 1.2 19. Panama ................ 1.0 20. Switzerland ............. 0.8 Other ................. 11.1 1.1% 0.9% 0.8% 0.7% "~:"~’" .6% MiTT is Minutes of Telecommunications Traffic. Data are in millions of minutes for public voice circuits. © TeleGeography, Inc. 1996 National Traffic Balance MiTT 1993 1994 1995 Incoming 148.3 164.3 186.6 Outgoing 133.3 141.3 129.1 Surplus (Deficit) 15.0 23.0 57.4 Total Volume 281.6 305.6 315.7 Nete: Data based on billing point of traffic. Totals may appear inconsistent with other figures due to rounding. TeleGeography 1996/97 © TeleGeography, Inc. 1996 Europe International Traffic TeleGeography 1996/97 © TeleGeography, Inc. 1996 Largest Destination MiTT 1. Germany ............. 344.5 2. Switzerland ........... 49.3 3. Italy .................. 42.5 4. Yugoslavia ............ 31.D 5. Hungary .............. 28.6 6. United States .......... 24.9 7. Turkey ................ 24.5 8. Croatia ............... 23.1 9. France ............... 21.0 10. Poland ............... 20.3 11. Netherlands ........... 19,4 12. Czech Republic ........ 18.3 13. United Kingdom ........ 16.6 14. Slovenia .............. 14.4 15. Slovak Republic ........ 11.0 16. Romania ............... 9.1 17. Russia ................. 9.1 18. Belgium ............... 8.7 19. Sweden ............... 7.4 Other ................. 95.5 Telecommunications Routes, 1994 Percentage of Outgoing Traffic ~:~: 2.8% 2.6% ":" 2 5% i~~ 2,4O/o T 2.0% 1.8% 1.3% 11% ’1.1% 1.1% Mi]-l- is Minutes of Telecommunications Traffic. Data are in millions of minutes for public voice circuits. © TeleGeography, Inc. 1996 National Traffic Balance MiTT 1993 1994 1995 Incoming 751.0 774.5 n.a. Outgoing 767.4 819.2 901 Surplus (Deficit) (16.4) (44.7) n.a. Total Volume 1,518.4 1,593.7 n.a. © TeleGeography, Inc. 1996 International Telephone Traffic--Europe Belgium Destination Largest Telecommunications Routes, 1. France .............. 263.0 2. Netherlands .......... 229.8 3. Germany ............. 137.4 4. United Kingdom ........ 97.0 5. Italy .................. 58.0 1995 Percentage of Outgoing Traffic ,,. ..,,,., "cs.,..Q~, ...... ’~. -. -:; ,.,,u", ~. ,- ,.’. ~,-;÷. 23.8% ......... 6. United States .39.9 ~;~,o~,,/3.6% 7. Luxembourg ........... 39.6 8. Spain ................ 35.6 ~!],~i~;.," 3.2% 9. Switzerland ........... 25.4 i!iil]~i;~,,, 2.3% 10. Sweden .............. 12.7 11. Portugal .............. 11,8 12. Greece ............... 11,0 i~i’i 1.o% 13. Denmark .............. 10.3 ~,!!i" 0.9% 14. Turkey ................ 10.2 !~i; 0.9% 15. Austria ................ 9.7 if!is o.9% 16. Morocco .............. 9.3 i-,~ o.9% 17. Poland ................ 8.7 i;i~ o.8% 18. Ireland ................ 6.2 ~]!i 0.6% 19. Canada ................ 5.5 ; 0.5% 20. Norway ............... 5.0 o.4% Other ................. 79.6 MiTT is Minutes of Telecornmunications Traffic. Data are in millions of minutes for public voice circuits. © TeleGeography, Inc. 1996 National Traffic Balance MiTT 1993 1994 1995 Incoming 1,025.3 1,093.9 1,172.0 Outgoing 979.4 1,049.0 1,105.7 Surplus (Deficit) 45.9 44.9 66.3 Total Volume 2,004.7 2,142.9 2,277.7 Note: Data based on billing point of traffic. Totals may appear inconsistent with other figures due to rounding. TeleGeography 1996/97 © TeleGeography, Inc. 1996 Croatia Destination Largest Telecommunications Routes, 1. Germany .............. 47.6 2. Bosnia ................ 46.7 3. Slovenia .............. 27.0 4. Italy .................. 17.2 5. Austria ............... 14.4 1995 Percentage of Outgoing Traffic ~’~ 22.6% 22.2% 6. Switzerland ............ 6.3 7. United Kingdom ......... 5.7 8. United States ........... 5.2 9. France ................ 4.3 10. Netherlands ............ 3.3 11. Canada ................ 2.6 12. Sweden ............... 2.4 13. Macedonia, TFYR ........ 2.3 14. Hungary ............... 2.2 15. Belgium ............... 2.1 16. Spain ................. 2.1 17. Australia .............. 1.8 18. Russia ................. 1.6 19. Czech Republic ......... 1.6 20. Denmark .............. 1.1 Other ................. 13.1 8.2% 6.8% 2.7% l~ 2.0% ~iil 1.6% 1.2% 1.1% ~i 1.1 ~/o 1.o% 1.o% o.9% o.8% MiTI is Minutes of Telecommunications Traffic. Bata are in millions of minutes for public voice circuits. © TeleGeography, Inc. 1996 National Traffic Balance MiTT 1993 1994 1995 Incoming 170.3 240.2 309.0 Outgoing 117.2 185.5 210.7 Surplus (Deficit) 53.0 54.8 98.3 Total Volume 287.5 425.7 519.7 Notes: Data based on billing point of traffic. 1993 totals do not include traffic to and from Bosnia. Totals may appear inconsistent with other figures due to rounding. © TeleGeographyo Inc. 1996 International Telephone Traffic--Europe Cyprus Destination Largest Telecommunications Routes, MiTt 1. United Kingdom ........ 30.9 2. Greece ............... 26.0 3. United States ........... 5.7 4. Russia ................. 5.4 5. Germany .............. 5.0 1995 Percentage of Outgoing Traffic 6. Romania ............... 2.9 7. Lebanon ............... 2.8 8. Italy ................... 2.5 9. Yugoslavia ............. 2.4 10. France ................ 2.3 11. Syria .................. 2.1 12. Bulgaria ............... 2.0 13. Egypt ................. 1.7 14. Switzerland ............ 1.6 15. Canada ................ 1.6 16. Ukraine ................ 1.5 17. Netherlands ............ 1.4 18. Israel ................. 1.3 19. Sweden ............... 1.3 20. Austria ................ 1.0 Other ................. 16.1 MiTT is Minutes of Telecommunications Traffic. Data are in millions of minutes for public voice circuits. @ TeleGeography, Inc. 1996 National Traffic Balance MiTT 1993 1994 1995 Incoming 72.2 79.0 87.3 Outgoing 93.8 106.6 117.4 Surplus (Deficit) (21.6) (27.5) (30.2) Total Volume 166.0 185.6 204.7 Note: Totals may appear inconsistent with other figures due to rounding. TeleGeography 1996/97 h Rep bli ,, zec u c ’ © TeleGeography, Inc. 1996 Destination Largest Mii~" Telecommunications Routes, 1995 Percentage of Outgoing Traffic 1. Germany .............. 58.2 2. Austria ............... 16.6 3. United Kingdom ........ 11.4 4. Italy ................... 9.3 5. France ................ 7.6 6. Poland ................ 7.3 7. United States ........... 7.3 8. Netherlands ............ 5.9 9. Russia ................. 5.7 10. Switzerland ............ 5.6 3.0% 11. Ukraine ................ 4.6 ~i~il 2.5% 12. Vietnam ............... 3.6 1.9% 13. Belgium ............... 3.5 ~!i 1.9% 14. Canada ................ 3.1 ~!ii’! 1.6% 15. Israel .................. 2.9 1.5% 16. Hungary ............... 2.7 ~! 1.4% 17. Spain ................. 2.5 ~ii’~ 1.3% 18. Sweden ............... 2.2 ~ii;i 1.2% 19. Yugoslavia ............. 2.1 i~: 1.1% 20. Croatia ................ 1.8 ~i 1.0% Other ................. 23.1 MiTT is Minutes of Telecommunications Traffic. Bata are in millions of minutes for public voice circuits. © TeleGeography, Inc. 1996 National Traffic Balance MiTT 1993 1994 1995 Incoming n.a. 210.0 223.7 Outg oing 141.4 157.6 186.8 Surplus (Deficit) n.a. 52.4 36.9 Total Volume n.a. 367.6 410.5 Nete: Data based on billing point of traffic and exclude traffic to and from the Slovak Republic. © TeleGeography, Inc. 1996 International Telephone Traffic--Europe Denmark Destination Largest Telecommunications Routes, MiTT 1. Germany ............. 102.6 2. Sweden .............. 88.3 3. United Kingdom ........ 56.8 4. Norway .............. 52.0 5. United States .......... 26.5 1995 Percentage of Outgoing Traffic 19.3% 6. France ............... 22.9 7. Netherlands ........... 21.9 8. Italy .................. 12.9 9. Finland ............... 10.9 10. Switzerland ........... 10.7 11. Belgium .............. 10.7 12. Spain ................ 10.1 13. Faroe Islands ........... 9.2 14. Poland ................ 8.8 15. Turkey ................. 5.8 16. Greenland .............. 4.8 17. Austria ................ 4.6 18. Iceland ................ 4.0 19. Canada ................ 3.8 20. Greece ................ 3.4 Other ................. 61.9 2.4% MiTT is Minutes of Telecommunications Traffic. 0.9% 0.6% Data are in millions of minutes for public voice circuits. @ TeleGeography, Inc. 1996 National Traffic Balance MiTT 1993 1994 1995 Incoming 460.0 500.9 551.0 0 utg o i n g 452.3 488.4 532.6 Surplus (D efi cit) 7.7 12.4 18.4 Total Volume 912.3 989.3 1,083.6 Note: Totals may appear inconsistent with other figures due to rounding. TeleGeography 1996/97 © TeleGeographyo Inc. 1996 Estonia Destination 1. Finland 2. Russia ................ 13.9 3. Sweden ............... 4.3 4. Germany .............. 2.9 5. Ukraine ................ 2.5 6. Latvia .2.5 ................. ~!i~)~!:~,,~;~,:: 4.7% 7. Lithuania ............... 1.6 3.o% 8. United States ........... 1.4 ~i 2.6% 9. Belarus ................ 1.1 10. Denmark .............. 1.0 1.9% 11. United Kingdom ......... 1.0 1.9% 12. Netherlands ............ 0.6 ~!:,~:! 1.2% 13. Norway ............... 0.5 ~!: O.9% 14. France ................ 0.4 ii~, 0.8% 15. Italy ................... 0.3 o.6% 16, Poland ................ 0,3 i~i o,6% 17. Belgium ............... 0.3 !i~0.6% 18. Switzerland ............ 0.3 i.! 0.5% 19. Canada ................ 0.2 ~, 0.4% 20. Kazakhstan ............. 0.2 o.4% Other ................. 2.1 ~!i~ 3.9% Largest Telecommunications Routes, MiTT Percentage of Outgoing Traffic ............... 15,6 1995 Mi’l-I- is Minutes of Telecommunications Traffic. Data are in millions of minutes for public voice circuits. © TeleGeography, Inc. 1996 National Traffic Balance MiTT 1993 1994 1995 Incoming n.a. 50.8 56.0 Outgoing 41.2 48.1 53.0 Surplus (Deficit) n.a. 2.7 3.0 Total Volume n.a. 98.9 109.0 ~ote: Data based on billing point of traffic. Totals may appear inconsistent with other figures due to rounding. © TeleGeography, Inc. 1996 International Telephone Traffic--Europe Finland Destination Largest Telecommunications Routes, Min" 1. Sweden .............. 99.0 2. Germany .............. 31.5 3. Russia ................ 24.0 4. United Kingdom ........ 21.1 5. Estonia ............... 18.3 6. United States .......... 15.8 7. Norway .............. 11.4 8. Denmark .............. 10.4 9. France ................ 9.1 10. Netherlands ............ 7.9 11. Switzerland ............. 7.2 12. Italy ................... 5.3 13. Belgium ................ 5.2 14. Spain .................. 4.5 15. Canada ................ 3.1 16. Turkey ................. 1.7 17. Austria ................. 1.7 18. Poland ................. 1.6 19. Hungary ............... 1.4 20. Greece ................ 1.3 Other ................. 33.9 ~;!i 3.3% ~i 2.9% ~iii, i 2.5% ~i~ 2.3% ~;~i 1.6% ;~" 1.4% !~i,. 1.0% ’ 0.5% 0.5% 0.5% 0.4% 0.4% MiTT is Minutes of Telecommunications Traffic. Data are in millions of minutes for public voice circuits. © TeleGeography, Inc. 1996 National Traffic Balance MiTT 1993 1994 1995 Incoming n.a. 285 345.0 Outgoing n.a. 259 315.4 Surplus (Deficit) n.a. 26 29.6 Total Volume n.a. 544 660.4 Nete: Data are rounded to the nearest million minutes. Data include Telecom Finland, Finnet International, Telivo, ~lands Mobiltelefon, HTC and Botnia Link Ltd. © TeleGeography. Inc. 1996 Largest Telecommunications Routes, 1995 Destination MiTI" 1. Germany ............. 325.9 2. United Kingdom ....... 317.7 3. Italy ................. 242.1 4. Belgium ............. 228.2 5. Spain ............... 179.7 6. United States ......... 165.8 7. Switzerland .......... 154.6 8. Portugal ............. 131.9 9. Netherlands ........... 97.2 10. Morocco ............. 88.5 11. Percentage of Outgoin0 Traffic 6.4% 4.7% Algeria ............... 67.6 12. Tunisia ................ 48.0 13. Canada ............... 37.7 14. Turkey ................ 33.7 15. Sweden ............... 26.6 16. Poland ................ 24.5 17. Denmark .............. 22,4 18. Austria ................ 20.8 19. Greece ............... 20.5 20. Luxembourg ........... 19.8 Other ................ 551.4 ~,~,, 1.3% 0.9% 0.9% 0.8% ¯ 0.7% O.7% ~:,: 0.7% 19.7% MiTT is Minutes of Telecommunications Traffic. Data are in millions of minutes for public voice circuits. © TeleGeography, Inc. 1996 National Traffic Balance MiTT 1993 1994 1995 Incoming 2,710.0 2,739.5 2,958.9 Outgoing 2,576.0 2,602.5 2,804.6 Surplus (Deficit) 134.0 137.0 154.3 Total Volume 5,286.0 5,342.0 5,763.5 Note: Totals may appear inconsistent with other figures due to rounding. 1OO © TeleGeography, Inc. 1996 International Telephone Traffic--Europe Germany Largest Telecommunications Routes, 1995 Destination MiTT Percentage of Outgoing Traffic 1. Austria .............. 416.3 2. France .............. 389.8 3. Switzerland .383.0 ~>~ik~-~,:P%;~;!~!~; 7.3% 4. Turkey .375.6 ~’,,:~:,;)>d~;:>~>~b~q~’~: 7.2% 5. Italy ................. 375.5 ~,~;~,~,~;,: 7.2% 6. United Kingdom .365.1 ~#’4~.~.:~: ~>~>~>s 7.0% 7. Netherlands .......... 335.9 8. United States ......... 283.6 ~;~h;~:?>’,:~;:>:~,L~,:~~: 5.4% 9. Poland .............. 278.9 ~J~,&~;p>,i’~v~x:sT;;~s#:" 5.3% 10. Spain ............... 170.4 11. Belgium ............. 1~.4 12. 6reece .............. 123.5 13. Croatia .............. 113.3 2.2% 14. Benmark ............. 103.5 ~]:~]~]~ 2.0% 15. Czech Republic ........ 86.5 ~:’~]*~/} 1.6% 16. Sweden .............. 77.3 ~’~ 1.5% 17. Yugoslavia ............. 77.1 ~,]’~]~, 1.5% 18. Russia ................ 71.7 ~]~:;~, 1.4% 19. Po~ugal .............. 67.7 ~;,~;S, 1.3% 20. HungaH .65.7 ............. Other .939.2 ................................................................ 17.9% MiTT is Minutes of Telecommunications Traffic. Data are in millions of minutes for public voice circuits. © TeleGeography, Inc. 1996 National Traffic Balance MiTT 1993 1994 1995 Incoming 3,707.8 3,881.2 n.a. 0 utg o in g 4,679.6 5,147.1 5,244.0 Surplus (Deficit) (971.8) (1,265.9) n.a. Total Volume 8,387.4 9,028.3 n.a. Note: Data are for Deutsche Telekom only. From January to June 1996, Deutsche Telekom reported total outgoing traffic of 2,256 million minutes and total incoming of 2,146 million minutes. 101 TeleGeo{~raphy 1996/97 © TeleGeography, Inc. 1996 Greece Destination Largest 1. Germany .............. 82.5 2. United Kingdom ........ 68.6 3. Italy .................. 40.2 4. United States .......... 30.5 5. France ............... 23.0 Telecommunications Routes, 1995 Percentage of Outgoing Traffic 4.9% 6. Cyprus ............... 18.0 7. Canada ............... 16.9 18. Russia ................. 6.7 ~.4% 19. Austria ................ 6.5 20. Poland ................ 5.8 ~’.:. ~.2% Other ................. 75.7 MiTF is Minutes of Telecommunications Traffic. Data are in millions of minutes for public voice circuits. © TeleGeography, Inc. 1996 National Traffic Balance MiTT 1993 1994 1995 Incoming 406.1 441.2 505.4 0 utg o in g 336.2 422.7 467.9 Surplus (Deficit) 70.0 18.6 37.4 Total Volume 742.3 863.9 973.3 Note: Totals may appear inconsistent with other figures due to rounding. 102 © TeleCeography, Inc. 1996 International Telephone Traffic--Europe Hungary Destination Largest Mil-r 1. Germany .............. 59.4 2. Austria ............... 30.4 3. Romania .............. 14.4 4. Italy .................. 13.5 5. United Kingdom ........ 13.3 Telecommunications Routes, 1995 Percentage of Outgoing Traffic 5.8% 3.4% 2.5% 6. United States .......... 11.2 7. Yugoslavia ............. 8.4 8. France ................ 8.4 9. Russia ................. 7.5 10. Switzerland ............ 7.1 11. Netherlands ............ 6.6 12. Slovak Republic ........ 6.3 13. Ukraine ................ 5.5 14. Belgium ............... 3.8 ’ 1.5% 15. Sweden ............... 3.5 i’°~ 1.4O/o 16. Israel .................. 3.4 ;:I,1.4O/o 17. Poland ................ 3.3 ~,,i; 1.3% 18. Croatia ................ 3.2 i,,!- 1.3% 19. Czech Republic ......... 2.9 ~i:’i. 1.2% 20. Greece ................ 2.7 ~:i, 1.1% Other .32.5 ~,, :~-:’, ~,",,~::~?:~::::%g:~:,~,~:’~:,-~, 13.1% MiTT is Minutes of Telecommunications Traffic. Data are in millions of minutes for public voice circuits. © TeleGeography, Inc. 1996 National Traffic Balance MiTT 1993 1994 1995 Incoming 192.8 211.9 243.7 Outgoing 213.2 236.6 247.5 Surplus (Deficit) (20.4) (24.7) (3.8) Total Volume 406.0 448.5 491.2 Note: Data based on billing point of traffic. Totals may appear inconsistent with other figures due to rounding. 103 TeleGeography 1996/97 © TeleGeographyo Inc. 1996 Destination Largest 1. United States ........... 6.8 2. Denmark ............... 4.4 3. Sweden ................ 3.0 4. Norway ................ 2.5 5. Germany ............... 2.3 Telecommunications Routes, 1995 Percentage of Outgoing Traffic 6. United Kingdom ......... 2.2 7. France ................. 0.8 8. Netherlands ............ 0.7 9. Farce Islands ........... 0.6 10. Spain .................. 0.4 11 Canada .0.4 ~’ ’,, "~ 1.4% ................ 12. Italy ................... 0.4 1.4% 13. Belgium ................ 0.4 ~,~;~,: 1.4% 14. Finland .0.3 ................ ~!~,;~ 1.2% 15. Switzerland ............. 0.3 i!i!!!i 1.1% 16. Portugal ............... 0.2 ~,; o.6% 17. Austria ................. 0.2 0.6°/o 18. Russia ................. 0.1 !~,i 0.5% 19. Australia ............... 0.1 :,i! 0.4% 20. Ireland ................. 0.1 ~-~ 0.3% 7.6% 2.7% ................. @~,~....,. -. _. ,.~.~.~.~ MiT[ is Minutes of Telecommunications Traffic. Data are in millions of minutes for public voice circuits. ©Tele6eography, inc. 1996 National Traffic Balance Mi]-I- 1993 1994 1995 Incoming 23.4 25.5 28.4 Outgoing 24.1 26.0 28.9 Surplus (Deficit) (0.7) (0.4) (0.6) Total Volume 47.5 51.5 57.3 ~ute: Data based on billing point of traffic. Totals may appear inconsistent with other figures due to rounding. © TeleGeography, Inc. 1996 International Telephone Traffic--Europe Ireland Largest Telecommunications Routes, FY 1995/96 Destination MiTT Percentage of Outgoing Traffic 2. United States .......... 49 ~:~ 12.o% 3. Germany .............. 18 ~ii! 4.4% 4. France .14 ~~;’°~ 3.4% .............. 5. Netherlands ............ i!~’i 2.2%° 6. Italy ................... 7 !~i ~’7°/° 7. Spain ................. i.i~, 1.5% 8. Belgium ............... 5 !!! 1.2% 9. Canada ................ 5 ~! 1.2% 10. Australia .............. 4 i ~.0% 11. Switzerland ............ 3 if’ o-7% 12. Denmark .............. 2 :~ o.5% 13. Sweden ............... 2 0.5%° Oth e r ................. 18 i .~ii: 4.4% MiTt is Minutes of Telecommunications Traffic. Data are in millions of minutes for public voice circuits. © TeleGeography, Inc. 1996 National Traffic Balance MiTT FY 1993/94 1994/95 1995/96 Incoming 423.0 442.9 n.a. Outgoing 315.8 323.7 407 Surplus (Deficit) 107.2 119.2 n.a. Total Volume 738.8 766.5 n.a. Netes: Data rounded to the nearest million minutes. Traffic to Northern Ireland is excluded in both totals and route data. Totals may appear inconsistent with other figures due to rounding. Fiscal year ends 31 March. 105 TeleGeography 1996/97 © TeleGeography, Inc. 1996 Destination Largest 1. Germany ............. 299.0 2. France .............. 239.0 3. Switzerland .......... 181.8 4. United States ......... 154.1 5. United Kingdom ....... 149.3 Telecommunications Routes, 1995 Percentage of Outgoing Traffic 12.5% 8.1% 6. Spain .66.9 ~’ :~-,,,,~,~,~,, ~,,~ 3.5% ............... 7. Belgium .............. 54.7 2.9% 8. Austria ............... 49.0 ~:_~:~ 2.6% 9. Netherlands ........... 43.8 ~’:~:~ 2.3% 10. Yugoslavia ............. 36.7 ~,’?.:~:; 1.9% 11. Chile .................. 35.9 12. Greece ............... 32.6 ;’~;;~ 1.7% 13. Morocco ............. 31.2 :~; 1.6% 14. Romania .............. 29.6 1.6% 15. Poland ............... 29.0 ~~’~; 1.5% 16. Croatia ............... 26.2 ;~~. 1.4% 17. Tunisia ............... 24.1 ~,~ 1.3% 18. Canada ............... 22.0 ::~, 1.2% 19. Brazil ................. 18.2 ;,;; 1.0% 20. Sweden .............. 16.4 ,,~,~ 0.9% Other .368.8 ................................................................... 19.3% MiTT is Minutes of Telecommunications Traffic. Data are in millions of minutes for public voice circuits. © TeleGeography, Inc. 1996 National Traffic Balance MiTT 1993 1994 1995 Incoming 1,672.7 1,864.0 1,999.8 Outgoing 1,609.7 1,708.0 1,908.2 Surplus (Deficit) 63.0 156.0 91.6 Total Volume 3,282.4 3,572.0 3,908.1 Note: Totals may appear inconsistent with other figures due to rounding. 106 © TeleGeographyo Inc. 1996 International Telephone Traffic--Europe Luxembourg Destination Largest Telecommunications Routes, MiTT 1. Belgium .............. 52.6 2. Germany .............. 48.0 3. France ............... 46.4 4. Portugal .............. 14.9 5. United Kingdom ........ 13.3 1995 Percentage of Outgoing Traffic 6. Italy .................. 11.1 ~,~’~’~’;~:~,~:~ 4.8% 7. Netherlands ............ 8.2 8. Switzerland .6.4 ~;%T? ........... ,~,:4~,~;~,~,, 2.7% 9. United States ........... 5.5 2.4% 10. Spain ................. 3.4 ~i~! ~.5% 11. Denmark .............. 3.2 12. Austria ................ 2.1 0.9% 13. Sweden ............... 2.0 ~i!: 0.9% 14. Greece ................ 1.5 ~:~ 0.0% 15. Ireland ................ 1,0 !~ o.4% 16. Finland ................ 0.8 i; o.3% 17. Poland ................. 0.7 ’ o.3% 18. Japan ................. 0.7 ~i o.3% 19. Russia ................. 0.7 ~,, 0.3% 20. Canada ................ 0.6 ii o.3% Other .9.1 .............. ...... 3.9% ............... ~:~ MiTT is Minutes of Telecommunications Traffic. Data are in millions of minutes for public voice circuits. @ TeleGeography, Inc. 1996 National Traffic Balance MiTT 1993 1994 1995 Incoming 131.7 145.2 174.5 0 utgoing 199.3 213.5 232.2 Surplus (Deficit) (67.6) (68.3) (57.7) Total Volume 331.0 358.7 406.7 Nnte: Totals may appear inconsistent with other figures due to rounding. 107 TeleGeography 1996/97 © TeleGeography, Inc. 1996 Macedonia TFYR Largest Telecommunications Routes, 1995 Destination MiTT 1. Germany .............. 5.5 2. Bulgaria ............... 3.4 3. Croatia ................ 2.8 4. Switzerland ............ 2.4 5. Slovenia ............... 2.4 Percentage of Outgoing Traffic 6. Turkey ................. 2.1 7. United States ........... 2.1 8. Austria ................ 1.5 9. Italy ................... 1.5 10. Greece ................ 1.2 ~,~:,:~S~,_;_L 3.4% 11. Australia .............. 0.9 L:’:A%, "~:~ 2.5% 12. United Kingdom ......... 0.9 i!~,~5~!_,,~,i]i 2.51 13. France ................ 0.7 14. Sweden .0.6 ’,’ 1.7% 15. Netherlands ............ 0.6 ~~ 1.7% 16. Russia ................. 0.5 ~/.~’ ~.4% 17. Albania ................ 0.4 .~/, 1.1% 18. Canada ................ 0.4 ~~ 1.1% 19. Belgium ............... 0.3 ;" 0.t% 20. Czech Republic ......... 0.3 7,; 0.1% MiTT is Minutes of Telecommunications Traffic. Data are in millions of minutes for public voice circuits. © TeleGeography, Inc. 1996 National Traffic Balance MiTT 1993 1994 1995 Incoming 48.0 78.3 74.0 Outgoing 27.6 35.1 35.5 Surplus (Deficit) 20.4 43.2 38.5 Total Volume 75.6 113.4 109.5 Note: Data based on billing point of traffic. Totals may appear inconsistent with other figures due to rounding. 108 © TeleGeography, Inc. 1996 International Telephone Traffic--Europe Netherlands Destination Largest Telecommunications Routes, Mi’n 1. Germany ............. 339.5 2. Belgium ............. 238.6 3. United Kingdom ....... 179.9 4. France .............. 112.1 5. United States .......... 90.0 6. Italy .................. 46.6 7. Spain ................ 42.6 8. Switzerland ........... 41.0 9. Turkey ................ 30.0 10. Sweden .............. 25.3 11. Denmark .............. 22.3 12. Austria ............... 19.0 13. Canada ............... 16.5 14. Poland ............... 14.0 15. Norway .............. 12.8 16. Greece ............... 12.0 17. Portugal .............. 10.9 18. Morocco ............. 10.4 19. Ireland ............... 10.2 20. Russia ................. 9.0 Other ................ 175.9 1995 Percentage of Outgoing Traffic 03% E 2.8,/o 2.1% 1.3% 1.0% 0.9% 0.8% 0.7% 0.7% 0.7% 0.6% MiTT is Minutes of Telecommunications Traffic. Oata are in millions of minutes for public voice circuits. © TeleGeography, Inc. 1996 National Traffic Balance MiTT 1993 1994 1995 Incoming 1,159.0 1,290.9 1,453.0 Outgoing 1,238.2 1,345.8 1,458.7 Surplus (Deficit) (79.2) (54.9) (5.7) Total Volume 2,397.2 2,636.7 2,911.7 Note: Data based on billing point of traffic. Totals may appear inconsistent with other figures due to rounding. TeleGeography 1996/97 © TeleGeography, Inc. 1996 Largest Telecommunications Routes, Destination MiTT 1. Sweden ............. 111 2. Denmark .............. 60 3. United Kingdom ........ 59 4. United States .......... 32 5. Germany .............. 24 6. France ............... 13 7. Netherlands ........... 12 8. Finland ............... 10 9, Spain ................. 8 10, Italy ................... 7 11. Switzerland ............ 6 12. Belgium ............... 6 13. Russia ................. 5 14. Poland ................ 5 15, Canada ................ 4 16. Turkey ................. 3 17. Austria ................. 2 18. Iceland ................ 2 19. Greece ................ 2 20. Portugal ............... 2 Other ................. 58 Percentage of Outgoing Traffic .........~ ......... 5.6% ~, 3.0% ~:j~ 2.8% ~L::i~2,, 2.3% :~,~ 1.9% j,~, 1.6% 1.4% 1.2% 1995 0.7% 0.5% 0.5% 0.5% 0.5% MiTT is Minutes of Telecommunications Traffic. Data are in millions of minutes for public voice circuits. © TeleGeography, Inc. 1996 National Traffic Balance MiTT 1993 1994 1995 Incoming 322.5 352.0 373.2 Outgoing 376.2 395.5 431.5 Surplus (Deficit) (53.7) (43.5) (58.3) Total Volume 698.7 747.5 804.7 Note: Data are rounded to the nearest million minutes and are based on billing point of traffic. © TeleGeographyo Inc. 1996 International Telephone Traffic--Europe m Poland Destination Largest Telecommunications Routes, MiTT Percentage of Outgoing Traffic 1. Germany ............. 146.4 2. United States .......... 30.3 3. United Kingdom ........ 20.6 4. France ............... 20.1 5. Italy .................. 19.2 1995 6. Austria ............... 13.2 7. Netherlands ........... 12.1 8. Sweden .............. 12.0 9. Russia ................ 11.7 10. Ukraine .10.1 ~~ .............. 2.7% 11. Canada ................ 8.0 i,i:,i~i~,~i 2.1% 12. Belgium .7.8 ...... .............. .,~>~" 2 0 ~/o 13. Czech Republic .6.6 i7~’~ 1.7% 14. Denmark .............. 6.4 ~,,i:~_ 1.7% 15. Switzerland ............ 5.5 1.4% 16. Belarus ................ 5.4 1.4% 17. Netherlands Antilles ..... 3.6 ;i’.°.9% 18. Spain ................. 3.6 ~’~ o.9% 19. Australia .............. 3.2 i 0.8% 20. Hungary ............... 2.8 o.7% ................. 32.9 ~, 8.6% Other MiTT is Minutes of Telecommunications Traffic. Data are in millions of minutes for public voice circuits. © TeleGeography, Inc. 1996 National Traffic Balance MiTT 1993 1994 1995 Incoming 431.5 643.8 649.3 0 utg o in g 272.7 356.6 381.4 Surplus (Deficit) 158.8 287.2 267.9 Total Volume 704.2 1,000.4 1,030.7 Note: Data based on billing point of traffic. Totals may appear inconsistent with other figures due to rounding. 111 TeleGeography 1996/97 © TeleGeography, Inc. 1996 Destination Largest Telecommunications Routes, 1. France ................ 61.6 2. Spain ................. 45.2 3. Germany .............. 33.9 4. United Kingdom ........ 31.3 5. Switzerland ............ 15.3 1995 Percentage of Outgoing Traffic 6. United States .......... 12.5 7. Brazil ................. 12.4 8. Italy .................. 11.7 9. Netherlands ............ 9.5 10. Belgium ................ 9.4 11. Angola ................. 6.4 12. Canada ................ 4.3 13. Sweden ................ 3.3 14. Luxembourg ............ 3.2 15. Cape Verde ............. 3.2 16. Guinea-Bissau .......... 3.1 17. Mozambique ............ 2.8 18. Denmark ............... 2.3 19. South Africa ............ 2.1 20. Austria ................. 1.7 o.6% Other .8.6 /,:~,~7~7,~,~ 3.0% ................ ~,~,.~~,>~1~;~ MiTT is Minutes of Telecommunications Traffic. Oata are in millions of minutes for public voice circuits. © Tele6eography, Inc. 1996 National Traffic Balance MiTT 1993 1994 1995 Incoming 438.2 467.8 525.0 Outgoing 232.6 262.4 283.9 Surplus (Deficit) 205.6 205.4 241.1 Total Volume 670.8 730.2 808.9 Notes: Totals are combined for Portugal Telecom, which handles traffic to Europe, and CPRM, which handles overseas traffic. In 1995 Portugal Telecom handled 238.1 million MiTT outgoing and 427.8 million MiTT incoming. Data for Portugal Telecom based on billing point of traffic. 112 © TeleGeography, Inc. 1996 International Telephone Traffic--Europe Russia Destination Largest Telecommunications Routes, 1995 Mi’l’r Percentage of Outgoing Traffic 1 Germany .............. 35.6 2. United States .......... 22.7 3. LaMa ................. 22.4 ~, ~,~, ~;~;,: ,4,~:4~-~;~, 7.8% 4. Lithuania .............. 16.9 5. United Kingdom ........ 14.0 6. Estonia ............... 13.9 7. Finland ................ 11.6 8. Italy .................. 11.5 9. France ................ 9.8 10. Turkey .9.4 ......... ........ : 3.3% ................ 11. Yugoslavia ............. 8.6 ~:~ 3.0% 12. Israel ................. 8.3 2.9% 13. Poland .7.5 ............... ~/~;;,~ ~,~,,, 2.6% 14. Austria ................ 5.2 4~:~’~ ~.8% 15. Switzerland ............ 5.2 ~:;~,7, 1.8% 16. Netherlands ............ 5.1 ~: 1.8% 17. China ................. 4.6 18. Hungary ............... 4.4 ~:~ 1.5% 19. Bulgaria ............... 4.3 ~:~, 1.5% 20. Czech Republic ......... 4.1 ~" 1.4% Other ................. 62.3 ~;~;~:~:~ ~,;,~:~:~;~:,;~:,~;:~:,~,,~:~:::~:~;~:, 21.7% MiTT is Minutes of Telecommunications Traffic. Data are in millions of minutes for public voice circuits. © Tele6eography, Inc. !996 National Traffic Balance MiTT 1993 1994 1995 Incoming 268.0 365.0 448.1 Outgoing 201.0 229.2 287.4 Surplus (Deficit) 67.0 135.8 160.7 Total Volume 469.0 594.2 735.5 Note: Data are for Rostelecom only and do not include traffic to and from other former Soviet republics (see page 122 for CIS statistics). TeleGeography 1996/97 © TeleGeography, Inc. 1996 Slovak Republic Largest Telecommunications Routes, Destination MiTT 1. Germany .............. 12.3 2. Austria ................ 9.2 3. Hungary ............... 5.3 4. Italy ................... 3.2 5. United States ........... 2.9 6. United Kingdom ......... 2.9 7. Poland ................ 2.2 8. Ukraine ................ 2.1 ,3.6% 9. Russia .2.1 ’,’~, ............ 3.6% ................ 10. Switzerland .2.0 ~,~,~,,~ ~’ 3.3% 11. France ................. 1.8 12. Netherlands .1.3 ~;~,~,,~-~: 2.2% ........... 13. Israel .................. 1.1 !~’~i: 1.8% 14. Belgium ............... 0.9 ~ii 1.5% 15. Croatia ................ 0.8 i~i!~i’: 1.4% 16. Canada ................ 0.8 1.4% 17. Yugoslavia .............. 0.7 ~ii 1.2% 18. Sweden ................ 0.5 0.6% 19. Spain .................. 0.5 ii!:il 0.8% 20. Slovenia ............... 0.4 ~i!" o.7% Other ................. 5.7 MiTT is Minutes of Telecommunications Traffic. Percentage of Outgoing Traffic 1995 Data are in millions of minutes for public voice circuits. © Tele6eography, Inc. 1996 National Traffic Balance MiTT 1993 1994 1995 Incoming 33.6 68.5 81.6 Outgoing 30.5 52.5 58.8 Surplus (Deficit) 3.1 16.0 22.8 Total Volume 64.1 121.0 140.4 Nate: Totals may appear inconsistent with other figures due to rounding. Data do not include traffic to and from the Czech Republic. 114 © TeleGeography, Inc. 1996 International Telephone Traffic--Europe Slovenia Largest Destination MiTT 1. Croatia ............... 27.4 2. Germany .............. 15.4 3. Austria ............... 11.0 4. Yugoslavia ............ 10.1 5. Italy ................... 9.8 6. Bosnia ................. 3.6 7. Switzerland ............ 2.5 8. Macedonia, TFYR ....... 2.4 9. United Kingdom ......... 2.1 10. France ................ 1.9 11. United States ........... 1.8 12. Russia ................. 1.7 13. Hungary ............... 1.3 14. Czech Republic ......... 1.1 15. Netherlands ............ 0.8 16. Sweden ............... 0.7 17. Belgium ............... 0.6 18. Canada ................ 0.5 19. Spain ................. 0.4 20. Poland ................. 0.4 Other ................. 4.8 Telecommunications Routes, 1995 Percentage of Outgoing Traffic 9.7% ~,~..~~ 2.4% ~:~i~ 1.9% ~." 13% ,0.8% ~,- 0.7% 0.6% 0.5% O.4% 0.4% ..... !.’:’ 4.8% MiTT is Minutes of Telecommunications Traffic. Data are in millions of minutes for public voice circuits. © Tele6eography, Inc. 1996 National Traffic Balance MiTT 1993 1994 1995 Incoming 55.8 83.2 121.2 Outgoing 62.8 90.6 100.6 Surplus (Deficit) (7.0) (7.4) 20.6 Total Volume 118.6 173.8 221.8 Note: Totals may appear inconsistent with other figures due to rounding. 115 TeleGeography 1996/97 © TeleGeographyo Inc. 1996 Destination Largest 1. Germany ............. 160.7 2. France .............. 158.2 3. United Kingdom ....... 147.0 4. Italy .................. 68.2 5. United States .......... 48.3 6. Portugal .............. 40.6 7. Switzerland ........... 36.1 8. Belgium .............. 35.4 9. Netherlands ........... 35.3 10. Morocco ............. 22.8 11. Andorra ............... 17.2 12. Argentina ............. 17.1 13. Sweden .............. 14.5 14. Chile .................. 12.9 15. Colombia ............. 10.5 16. Denmark .............. 9.5 17. Mexico ................ 8.9 18. Brazil ................. 8.8 19. Cuba .................. 8.7 20. Austria ................. 8.4 Oth e r ................ 155.8 Telecommunications Routes, 1995 Percentage of Outgoing Traffic 15.4% 6.7% .,~ 4.0% MiTT is Minutes of Telecommunications Traffic. Data are in millions of minutes for public voice circuits. © Tele6eography, Inc. 1996 National Traffic Balance MiTT 1993 1994 1995 Incoming 908.4 969.9 1,076.4 Outgoing 846.9 948.3 1,024.6 Surplus (Deficit) 61.5 21.6 51.8 Total Volume 1,755.3 1,918.2 2,101.0 Note: Totals may appear inconsistent with other figures due to rounding. © TeleGeography, Inc. 1996 International Telephone Traffic--Europe Sweden Largest Telecommunications Routes, Destination Mil"[ 1. Finland ............... 128 2. Norway .............. 124 3. Denmark .............. 96 4. Germany .............. 95 5. United Kingdorn ........ 76 6. United States .......... 62 7. France ................ 36 8. Netherlands ........... 26 9. Italy .................. 19 10. Poland ................ 19 11. Switzerland ............ 16 12. Belgium ............... 16 13. Spain ................. 14 14. Austria ................ 10 15. Yugoslavia .............. 9 Other ................ 154 1995 Percentage of Outgoing Traffic Mi]-r is Minutes of Telecommunications Traffic. Data are in millions of minutes for public voice circuits. © TeleGeography, Inc. 1996 National Traffic Balance MiTr 1993 1994 1995 Incoming 630 n.a. n.a. Outgoing 740 802 900 Surplus (Deficit) (110) n.a. n.a. Total Volume 1370 n.a. n.a. Note: Data are for Telia and Tele 2 only based on billing point of traffic rounded to the nearest million minutes. 117 TeleGeography 1996/97 © TeleGeography, Inc. 1996 Destination Largest Telecommunications Routes, 1. Germany ............. 408.2 2. France ............... 280.0 3. Italy ................. 252.8 4. United Kingdom ........ 98.5 5. United States .......... 79.7 6. Austria ................ 76.1 7. Portugal ............... 60.2 8. Spain ................. 55.1 9. Netherlands ........... 42.9 10. Yugoslavia ............. 41.3 11. Belgium ............... 29.8 12. Turkey ................ 28.5 13. Croatia ................ 19.1 14. Sweden ............... 18.9 15. Canada ............... 15.4 16. Greece ............... 13.6 17. Macedonia ............ 12.1 18. Denmark .............. 11.8 19. Brazil ................. 10.7 20. Russia ................ 10.6 Other ................ 213.0 1995 Percentage of Outgoing Traffic ~;,~.~ 4.5% ............ ~: 3,4% ~,,,.,.:: 3.1% 2.4% 2.3% 1.7% ~! 1.1% ~!i 1.1% 0.8% 0.7% 0.7% 0.6% 0.8% Mi-I-F is Minutes of Telecommunications Traffic. Data are in millions of minutes for public voice circuits. © TeleGeography, Inc. 1996 National Traffic Balance MiTT 1993 1994 1995 In c o rain g 1,258.7 1,353.0 1,439.3 Outgoing 1,572.0 1,649.3 1,778.4 Surplus (Deficit) (313.3) (296.3) (339.1) Total Volume 2,830.7 3,002.3 3,217.7 Note: All route data are rounded to the nearest million minutes and based on billing point of traffic. 118 © TeleGeography, Inc. 1996 International Telephone Traffic--Europe Turkey Largest Telecommunications Routes, 1995 Destination Mi’IT Percentage of Outgoing Traffic 1. Germany ............. 124.5 2. United Kingdom .30.8 3. United States .......... 20.6 5.5% 4. France ............... 20.1 ~:~ 5.4% 5. Netherlands ........... 15.8 ~i!~ 4.2% 6. Russia ................ 15.4 7. Italy .................. 11.2 3.o% 8. Switzerland ........... 11.1 9. Austria ................ 9.2 ~i~’:~: 2.5% 10. Romania ............... 8.0 ~i 2.1% 11. Belgium ............... 7.2 !~i!~ 1.9% 12. Bulgaria ............... 6.7 ~:~’ 1.8% 13. Greece .6.0 ~~ 1 6% ............... ~’~i,, 14. Israel ................. 5.7 !~: 1.5% 15. SaudiArabia ........... 5.6 if" 1.5% 16. Ukraine ................ 5.2 i~;, 1.4% 17. Iran ................... 5.1 ii~ 1.4% 18. Azerbaijan .............. 5.0 !i:~ 1.3% 19. Sweden ............... 4.0 ii!i 1.1% 20. Denmark .............. 3.0 ~:~0.8% ,~!~:~;;~, 14,3% MiTT is Minutes of Telecommunications Traffic. Data are in millions of minutes for public voice circuits. © TeleGeography, Inc. 1996 National Traffic Balance Mi~-F 1993 199~, 1995 Incoming 605.0 601.4 705.0 0 utg o in g 264.6 284.3 373.6 Surplus (Deficit) 340.4 317.1 331.5 Total Volume 869.6 885.8 1,078.6 Nnte: Totals may appear inconsistent with other figures due to rounding. 119 TeleGeography 1996/97 © TeleGeography, Inc~ 1996 Destination Largest Telecommunications Routes, FY 1995/96 Mill" Percentage of Outgoing Traffic 1. United States ......... 617.6 2. Ireland .............. 371.4 3. Germany ............. 364.4 4. France .............. 360.8 5. Italy ................. 188.0 ~,~,,Z,, 4.7% 6. Netherlands .173.5 ~,~ ,~,’:,~ 4.3% ......... 7. Spain ............... 171.1 8. Australia ............. 127.3 II~,L.-.~],,~,,]~.] 3.2 9. Canada .............. 120.5 10. Belgium ............. 105.4 11. Swi~erland .102.9 ~,,:~;~:~, 2.5% 12. Sweden .............. 66.8 1.7% 13. Greece ............... 62.9 ,~:~ 1.6% 14. Denmark .............. 57.6 ~:~L~,:~ 1.4% 15. South Africa ........... 57.5 ~,~ 1.4% 17. India ................. 55.5 1.4% 18. Pakistan .............. 51.0 1.3% 19. Hong Kong ............ 51.0 1.3% MiTT is Minutes of Telecommunications Traffic. Data are in millions of minutes for public voice circuits. © Telel3eography, Inc. 1996 National Traffic Balance M iTF Pi’ 1993/94 I~’ 1994/95 FY 1995/96 Incoming 3086 3577 4,021 Outgoing 3130 3507 4,016 Surplus (Deficit) (44) 70 5 Total Volume 6216 7084 8037 Note: Data are for BT and Mercury only and exclude IPL reseller traffic as well as traffic between the Irish Republic and Northern Ireland. IPL resellers had approximately 280 million MWI of outgoing traffic in FY 1995/96, of which WorldCom, ACC and Sprint accounted for 240 million Mi’I-F. Data for carrier call minutes published in the Market Information Upate by the U.K. Office of Telecommunications (Oftel) may not match TeleGeography data because Oftel reports "retail" minutes only, which exclude (a) "wholesale" minutes sold to switched resellers and (b) operator assisted and collect calls. 120 © TeleGeography, Inc. 1996 International Telephone Traffic--Europe F .............. 53.6 2. Austria ............... 23.9 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. Yugoslavia Largest Telecommunications Routes, MiTT Percentage of Outgoing Traffic Switzerland ........... 23.9 Italy .................. 10.8 Macedonia, TFYR ........ 9.7 France ................ 9.6 Slovenia ............... 8.5 Russia ................. 7.2 Hungary ............... 7.1 Sweden ............... 6.4 Greece ................ 6.2 United States ........... 5.9 Canada ................ 4.3 United Kingdom ......... 3.7 Netherlands ............ 3.2 Australia .............. 2.5 Bulgaria ............... 2.0 Cyprus ................ 2.0 Turkey ................. 1.9 Czech Republic ......... 1.6 Other ................. 18.8 1995 MiTr is Minutes of Telecommunications Traffic. 2.9% ~:"i~*~’:,’ 2.0% iii" 1.2% 0.9% 0.9% 0.9% 0.8% Data are in millions of minutes for public voice circuits. @ Tele6eography, Inc. 1996 National Traffic Balance MiTT 1993 1994 1995 Incoming 223.5 229.0 296.0 Outgoing 181.5 181.9 212.8 Surplus (Deficit) 42.0 47.1 83.2 Total Volume 405.0 410.9 508.8 Note: Totals may appear inconsistent with other figures due to rounding. 121 TeleGeography 1996/97 © TeleGeography, Inc. 1996 122 Asia, Middle East & Africa International Traffic TeleGeography 1996/97 © TeleGeography, Inc. 1996 © TeleGeography, Inc. 1996 International Telephone Traffic--Asia, Middle East, & Africa Australia Largest Telecommunications Routes, 1995 Destination MiTr 1. United Kingdom ....... 182 2. United States ......... 158 3. New Zealand ......... 142 4. Hang Kong ............ 40 5. Japan ................ 34 6. Singapore ............ 28 7. Canada ............... 25 8. Germany .............. 24 9. Italy .................. 20 10. Malaysia ............. 20 Other ................ 351 Percentage of Outgoing Traffic MiTT is Minutes of Telecommunications Traffic. Data are in millions of minutes for public voice circuits. © TeleBeagraphy, Inc. 1996 National Traffic Balance MiTT 1993 1994 1995 Incoming n.a. n.a. n.a. Outgoing 735 852 1024 Surplus (Deficit) n.a. n.a. n.a. Total Volume n.a. n.a. n.a. Note: Data are for Telstra, 0ptus and IPL rese!lers (for 1995 only) rounded to the nearest million minutes. 125 TeleGeography 1996/97 © TeleGeography, Inc. 1996 Destination Largest 1. India .................. 16.2 2. Saudi Arabia ........... 13.8 3. United Arab Emirates ...10.6 4. United Kingdom ......... 8.1 5. Kuwait ................. 4.7 Telecommunications Routes, 1995 Percentage of Outgoing Traffic 6. Egypt .4.3 ~;~’~::~;~:~L~!~’,~;,~;~,~, 4.9% ................. 7. United States ........... 4.1 ~:~ ~:~,, 4.7% 8. Pakistan ............... 4.1 9. Qatar .3.6 4.1% ................. 10. Oman .1.9 ................. &,~t~7~, 2.1% 11. Philippines ............. 1.6 12. Jordan ................. 1.2 13. Sri Lanka ............... 1.2 1.3% 14. Bangladesh ............. 1.0 ~,~;~ 1.1% 15. Iran ................... 0.9 ~:~, 1.o% 16. France ................. 0.8 0.9% 17. Morocco ............... 0.7 ~,~;/0.8% 18. Germany ............... 0.7 ~: 0.8% 19. Syria .................. 0.6 0.7% 20. Japan ................. 0.5 Mi~ is Minutes of Telecommunications Traffic. Data are in millions of minutes for public voice circuits. @ TeleGeography, Inc. ;996 National Traffic Balance MiTT 1993 1994 1995 Incoming n.a. n.a. 299.4 Outgoing 77.0 86.8 88.7 Surplus (Deficit) n.a. n.a. 210.6 Total Volume n.a. n.a. 388.1 Nute: Data based on billing point of traffic. 126 © TeleGeography, Inc. 1996 International Telephone Traffic--Asia, Middle East, &Africa Bangladesh Largest Telecommunications Routes, 1995 Destination MiTT 1. United Kingdom ......... 3.7 2. India .................. 3.4 3. United States ........... 2.9 4. Singapore ............. 1.7 5. Hong Kong ............. 1.6 ~i,~}~!~,~Vi~ 4.8"/. 6. SaudiArabia .1.3 ............... 3.9% .......... 7. Korea, Rep. of .......... 1.3 ~>~ ~,~:~,~ 3.9% 8. Japan .1.0 ................ 3.1% 9. Pakistan .1.0 ~.~;3., 10. Malaysia .............. 1.0 ~:~:~, 3.o% 11. United Arab Emirates .... 0.8 12. Germany .............. 0.6 13. China ................. 0.6 14. Italy ................... 0.5 16. Canada ................ 0.4 1 2% 17. Sri Lanka ............... 0.2 Other ................. 10.5 Percentage of Outgoing Traffic MiTT is Minutes of Telecommunications Traffic. Data are in millions of minutes for public voice circuits. © TeleGeography, Inc. 1996 National Traffic Balance MiTT 1993 1994 1995 Incoming 83.9 n.a. 122.1 Outgoing 17.2 22.1 33.0 Surplus (Deficit) 66.7 n.a. 89.1 Total Volume 101.1 n.a. 155.1 127 TeleGeography 1996/97 © TeleGeography, Inc. 1996 Largest Telecommunications Routes, Destination Mil’r 1. Hong Kong ........... 750.0 3. Taiwan ............... 115.0 2. Japan ................ 86.8 4. United States .......... 75.0 5. Korea, Rep ............. 45.7 6. Macau ................ 41.4 7. Singapore ............. 24.6 8. Australia .............. 13.5 9. Germany .............. 12.0 10. Canada ............... 10.7 11. United Kingdom ......... 7.2 12. France ................. 6.8 13. Thailand ............... 6.5 14. Russia ................. 6.2 15. Malaysia ............... 6.0 16. Italy ................... 5.3 17. Indonesia .............. 3.5 18. Philippines ............. 3.0 19. Netherlands ............ 2.5 20. New Zealand ........... 2.0 Other ................ 115.5 1995 Percentage of Outgoing Traffic 1.0% 0.9% 0.8% 0.5% 0.5% 0.5% 0.5% 0.4% 0.4% 0.3% 0.2% 0.2% 0.1% MiTT is Minutes of Telecommunications Traffic. Data are in millions of minutes for public voice circuits. © TeleGeography, Inc. 1996 National Traffic Balance MiTT 1993 Incoming n.a. Outgoing 1090 Surplus (Deficit) n.a. Total Volume n.a. 1994 1995 n.a. n.a. 1170 1339.1 n.a. n.a. n.a. n.a. Note: Totals may appear inconsistent with other figures due to rounding. 128 © TeleGeographyo Inc. 1996 International Telephone Traffic--Asia, Middle East, & Africa Hong Kong Largest Telecommunications Routes, FY 1995/96 Destination Mi’l-r Percentage of Outgoing Traffic 1. China ............... 913.6 2. United States ......... 101.5 6,o% 3. Taiwan ............... 84.6 ~ii~ 5.0% 4. Japan ................ 67.7 ~i~i 4.0% 5. United Kingdom ........ 67.7 ~i.i~i’i 4.0% 6. Canada .50.8 ~!’~!: 3.o% 7. Singapore ............ 50.8 ~i:i! 3.o% 8. Australia .............. 50.8 i!i~ 3.o% 9. Philippines ............ 50.8 ~:i~ 3.o% 10. Macau ................ 33.8 ~iil 2.o% Other ................ 219.9 ~i~i 13.o% MiTT is Minutes of Telecommunications Traffic. Data are in millions of minutes for public voice circuits. © TeleGeography, Inc. 1996 National Traffic Balance M iTT FY 1993/94 1994/95 1995/96 I n c o m in g 1,260.3 1,446.4 1,598.3 0 utg o in g 1,376.9 1,578.4 1,691.8 Surplus (Deficit) (116.5) (132.1) (93.5) Total Volume 2,637.2 3,024.8 3,290.2 Note: Data based on billing point of traffic. Route-by-route traffic volumes reflect reported data of Hong Kong Telecom which has been rounded to the nearest percent.. Totals may appear inconsistent with other figures due to rounding. Fiscal year ends 31 March. 129 TeleGeography 1996/97 © TeleGeography, Inc. 1996 Destination 1. Saudi Arabia ........... 69.8 2. United States .......... 49.9 3. United Arab Emirates ...32.7 4. United Kingdom ........ 25.5 5. Singapore ............. 14.3 6. Germany .............. 13.2 7. Kuwait ................. 9.4 8. Canada ................ 8.7 9. Oman .................. 7.9 10, Hang Kong ............. 7.6 11. Pakistan ............... 7.4 12. Japan ................. 6.9 13. France ................. 5.8 14. Australia ............... 5.6 15. Italy ................... 5.1 16. Sri Lanka ............... 4.6 17. Qatar .................. 3.9 18, Netherlands ............ 3.3 19. Russia ................. 3.3 20. Switzerland ............. 3,2 Largest Telecommunications Routes, FY 1995/96 MiTr Percentage of Outgoing Traffic 2.7% ~i )2!~i 2.3% 2.2% 2.0% 1.7% 1.5% 1.3% 1.2% 1.0% 1.0% 0.9% Other .53.3 ............................................. ........ 15.6% Mi~ is Minutes of Telecommunications #affic. Data are in millions of minutes for public voice circuits. @ Tele6eography, Inc. 1996 National Traffic Balance MiTT FY 1993/94 1994/95 1995/96 Incoming 441.0 615.0 805.4 Outgoing 283.9 314.0 341.4 Surplus (Deficit) 157.1 300.9 464.0 Total Volume 724.8 929.0 1146.8 Note: Data based on billing point of traffic. Outgoing totals and route data do not include calls to Bangladesh and Nepal. Fiscal year ends 31 March. Totals may appear inconsistent with other figures due to rounding. 130 © TeleGeography, Inc. 1996 International Telephone Traffic--Asia, Middle East, & Africa Indonesia Destination Largest Telecommunications Routes, 1995 MiTT Percentage of Outgoing Traffic 1. Singapore ............ 47.8 2. United States .......... 20.6 3. Japan ................ 19.9 4. Australia .............. 17.5 5. Malaysia .............. 13.3 6. Hong Kong ............ 12.4 7. Taiwan ................ 9.0 8. Korea, Rep. of ........... 8.8 9. United Kingdom ......... 6.7 10. Germany .............. 5.5 11. China .................. 4.6 12. Netherlands ............ 4.3 13. Philippines ............. 3.7 14. Saudi Arabia ............ 3.5 15. France ................ 3.3 16. Thailand ............... 3.1 17. India .................. 2.8 18. Italy ................... 2.0 19. Canada ................ 2.0 20. Switzerland ............ 1.3 Other ................. 13.8 MiTT is Minutes of Telecommunications Traffic. Data are in millions of minutes for public voice circuits. © Tele6eography, Inc. 1996 National Traffic Balance MiTT 1993 1994 1995 Incoming 201.8 244.7 286 Outgoing 143.8 182.5 205.9 Surplus (Deficit) 58.0 62.2 80 Total Volume 345.6 427.2 492 Note: Totals may appear inconsistent with other figures due to rounding. Data based on billing point of traffic. 131 TeleGeography 1996/97 © TeleGeography, Inc. 1996 Largest Telecommunications Routes, Percentage of Outgoing Traffic 1995 Destination MiTT 1. United Arab Emirates ...22.3 2. Kuwait ............... 22.3 3. Germany .............. 21.2 4. United States .......... 17.8 5. United Kingdom ........ 10.1 6. Pakistan ............... 8.8 7. Turkey ................. 7.6 8. Sweden ............... 6.8 9. Canada ................ 6.3 10. Japan ................. 6.1 11. France ................ 5.1 12. SaudiArabia ........... 3.8 ~i~ 1.8% 13. Italy ................... 3.6 @i ~.7%o 14. Netherlands ............ 3.3 1.6% 15. Qatar .................. 2.7 ~!~, ~.3% 16. Austria ................ 2.1 ~ii~: 1.o% 17. Switzerland ............ 2.1 ~i 1.o% 18. India .................. 1.8 !;i:::! 0.9% 19. Denmark .............. 1.4 I!i~ o.7% 20. Spain .................. 0.9 !i:, 0.4% MiTT is Minutes of Telecommunications Traffic. Data are in millions of minutes for public voice circuits. © Tele6eography, Inc. 1996 National Traffic Balance MiTT 1993 1994 1995 Incoming n.a. n.a. 199 Outgoing 156.5 208.4 210.4 Surplus (Deficit) n.a. n.a. (11) Total Volume n.a. n.a. 409 Note: Data based on billing point of traffic. © TeleGeography, Inc. 1996 International Telephone Traffic--Asia, Middle East, &Africa Israel Destination Largest Telecommunications Routes, 1995 MiTT Percentage of Outgoing Traffic 2. United Kingdom ........ 2~.9 3. France ............... 16.1 4. Jordan .15.1 ~)’~17: 8~::~’~"~’%~ 6.0% ............... 5. Germany .............. 14.0 s~~2~;~~"~’~ 5.6% 6. Russia ................ tl.7 7. Canada ................ 9.0 8. Italy ................... 7.9 9. Ukraine ................ 7.5 ~: 3.0% 10. Romania ............... 5.9 2.3% 11. Netherlands ............ 5.4 ~~:~, 2.1% 12. Swi~erland ............ 5.3 ~,~ 2.1% 13. Belgium ................ 4.3 ~", 1.7% 14. Turkey ................. 3.8 ~f:’ 1.5% 15. Egypt .................. 3.6 ~’’ ~.4% 16. South Africa ........... 3.0 ~.2% 17. Australia .............. 2.6 ~~~ 1,0% 18. Dominican Republic ...... 2.6 ~;~ 1.0% 19. Argentina .............. 2.6 .~ 20. Spain ................. 2.3 0.9% MiTT is Minutes of Telecommunications Traffic. Data are in millions of minutes for public voice circuits. © Tele6eography, Inc. 1996 National Traffic Balance MiTT 1993 1994 1995 Incoming 309.5 n.a. 345.6 Outgoing 175.5 213.0 252.3 Surplus (Deficit) 134.0 n.a. 93.3 Total Volume 485.0 n.a. 597.9 Nate: Data based on billing point of traffic. 133 TeleGeography 1996/97 © TeleGeography, Inc. 1996 Japan Destination Largest Telecommunications Routes, FY 1995/96 MiTT Percentage of Outgoing Traffic 1. United States ......... 348.9 2. China ................ 202.1 3. Korea ................ 155.7 4. Philippines ........... 139.0 5. Taiwan ................ 85.0 QQo/^ 13, Thailand .............. 68.8 7. Brazil ................. 60.4 ~:~i,~!~ 3.8% 8. Hong Kong ............ 57.8 9. United Kingdom ........ 48.0 ~i~’~’~i!i! 3.0% 10. Singapore .39.1 ~;:~,2.5% ............ 11. Australia .............. 33.4 12. Malaysia .............. 32.3 [~: 2.0% 13. Indonesia .27.2 ............ 14. Germany .............. 25.9 ~.~,~/ 1.6% 15. Canada ............... 23.1 ~/1.5% 16. France ................ 21.0 ~,~ 1.3% 17. Peru .................. 20.7 ;~;~ ~.3% 18. Iran .................. 14.7 ~: 0.9% 19. Pakistan .............. 12.5 ~,,;o.8% 20. Italy .................. 11.9 Mi~ is Minutes of Telecommunications Traffic. Data are in millions of minutes for public voice circuits. @ TeleGeography, Inc. 1996 National Traffic Balance Mi’FI FY 1993/94 FY 1994/95 FY 1995/96 Incoming 981.2 1140.6 1321 Outgoing 1411.2 1524.8 1631 Surplus (Deficit) (429.8) (384.2) (310) Total Volume 2392.4 2665.4 2952 Note: Route data include only IDD calls, while total data include operator assisted calls as well. Fiscal year ends 31 March. Data are for KDD, tTJ, and IDC combined. © TeleGeographyo Inc. 1996 International Telephone Traffic--Asia, Middle East, & Africa Jordan Destination 1. Israel ................. 9.8 2. Saudi Arabia ........... 8.6 3. Iraq ................... 7.9 4. Egypt ................. 6.4 5. Syria .................. 6.4 6. United States ........... 4.4 7. United Arab Emirates .... 4.1 8. United Kingdom ......... 2.9 9. Kuwait ................ 2.5 10. Lebanon ............... 2.4 11. Germany .............. 1.4 12. Italy ................... 1.1 13. Qatar .................. 1.0 14. France ................ 1.0 15. Oman ................. 0.8 16. Yemen ................. 0.8 17. Turkey ................. 0.7 18. Bahrain ................ 0.7 19. Canada ................ 0.6 20. Switzerland ............. 0.5 Other ................. 7.8 MiTT is Minutes of Telecommunications Traffic. Largest Telecommunications Routes, 1995 MiTT Percentage of Outgoing Traffic :~>’,’,~;~,’;~ ~-,~,~-’~’;~’-~’,~ 6.1% ~’~7~.$~(s~,~;"~’D",~,;+’2>,’;,1 4.0% ,~:;~,fi~:,4 1.9% ! 71.5%o ~?~i2~5 1.4% i! ii!ii,i 1"1°/° ~,,,. 1.0% L.? 0.8% Data are in millions of minutes for public voice circuits. © TeleGeography, Inc. 1996 National Traffic Balance MiTT 1993 1994 1995 Incoming 100 114 118.0 Outgoing 50 57 71.7 Surplus (Deficit) 50 57 46.3 Total Volume 150 171 189.7 Nntes: Traffic to Israel includes traffic to the West Bank. 135 TeleGeography 1996/97 © TeleGeography, Inc. 1996 Largest Telecommunications Routes, 1995 Destination MiTI" Percentage of Outgoing Traffic 1. United States .141 2. Japan ............... 126 6. Australia .12 ............ ~j~ 2.2% 7. Philippines ............ 12 ~!~i 2.2% 2.0% 10. Canada .10 ’W, 1.8% 11. Taiwan ................ 9 :,’,; 1.6% 12. Singapore .9 .... ’; 1.6% 13. France ................ 8 1.4% 15. Vietnam ................ 6 1.1% MiTT is Minutes of Telecommunications Traffic. Data are in millions of minutes for public voice circuits. © TeleGeography, Inc. 1996 National Traffic Balance MiTT 1993 1994 1995 Incoming 510.5 555.2 672 0 utg o in g 355.4 440.4 557 Surplus (Deficit) 155.1 114.8 115 Total Volume 865.9 995.6 1,229 Note: Data are for Korea Telecom and DACOM combined and are based on billing point of traffic. 136 © TeleGeography, Inc. 1996 International Telephone Traffic--Asia, Middle East, Africa Kuwait Largest Telecommunications Routes, Destination MiTr 1. Egypt ................. 21.1 2. Saudi Arabia .......... 14.4 3. India ................. 12.9 4. United Arab Emirates .... 8.1 5. Syria .................. 8.1 6. United States ........... 7.8 7. Pakistan ............... 7.5 8. United Kingdom ......... 7.2 9. Jordan ................ 5.8 10. Iran ................... 5.3 11. Bahrain ............... 3.4 12. Lebanon ............... 2.9 13. Bangladesh ............ 1.8 14. France ................ 1.3 15. Germany .............. 1.3 16. Qatar .................. 1.2 17. Philippines ............. 1.2 18. Canada ................ 1.1 19. 0man .................. !.0 20. Italy ................... 1.0 MiTT is Minutes of Telecommunications Traffic. Data are in millions of minutes for public voice circuits. © TeleGeography, Inc. 1996 National Traffic Balance MiTT 1993 1994 1995 Incoming n.a. 127.0 130.2 Outgoing 116.8 120.6 114.5 Surplus (Deficit) n.a. 6.4 15.7 Total Volume n.a. 247.6 244.7 Note: Totals may appear inconsistent with other figures due to rounding. Outgoing totals for 1995 excludes traffic to countries outside the top 20 routes, thus overstating the surplus. 137 TeleGeography 19913/97 © TeleGeography, Inc. 1996 Largest Destination MiTT 1. China ................ 49.0 2. Hong Kong ............ 45.4 3. Portugal ............... 3.1 4. Taiwan ................ 2.4 5. United States ........... 1.2 6. Thailand ............... 1.1 7. Canada ................ 1.0 8. Philippines ............. 0.8 9. Australia .............. 0.6 10. United Kingdom ......... 0.5 11. Singapore ............. 0.4 12. Japan ................. 0.4 13. Malaysia .............. 0.3 14. France ................ 0.2 15. Korea, Rep. of .......... 0.2 16. Indonesia .............. 0.2 17. Vietnam ................ 0.1 18. Germany ............... 0.1 19. Brazil .................. 0.1 Other .................. 1.1 Telecommunications Routes, 1995 Percentage of Outgoing Traffic ~;~: 2.2% ~!~ 1 .o% 0.9% ~, 0.8% 0.5% o.4% 0.4% 0.4% 0.3% o.2% 0.2% 0.1% o.1% o.1% o,1% ~,I 1.1% MiTT is Minutes of Telecommunications Traffic. Data are in millions of minutes for public voice circuits. © TeleGeography, Inc. 1996 National Traffic Balance MiTT 1993 1994 1995 Incoming 78.0 84.3 90.4 Outgoing 89.9 100.0 108.1 Surplus (D eficit) (11.9) (15.7) (17.7) Total Volume 167.9 184.3 198.5 Note: Totals may appear inconsistent with other figures due to rounding. Data based on billing point of traffic. 138 © TeleGeography, Inc. 1996 International Telephone Traffic--Asia, Middle East, & Africa Malaysia Largest Telecommunications Routes, FY 1995/96 Destination Mi’l’r 1. Singapore ............ 184.0 2. Japan ................ 26.0 3. United Kingdom ........ 22.3 4. Indonesia .21.1 ;~:~:,~ 5.2% ............ 5. Australia .............. 19.9 ~! 4.9% 6. United States .......... 17.2 7. Hong Kong ............ 15.8 ~i, 3.9% 8. Thailand .............. 12.9 ~!, 3.2% 9. Taiwan ............... 12.0 10. India .................. 9.3 ~ii! 2.3% 11. Philippines ............. 9.3 ~!:i~ 2.3% 12. China ................. 7.6 ~!~,i!; 1.9% 13. Bangladesh ............. 5.2 ~i:~ 1.3% 14. Germany .............. 5.1 i!i! 1.2% 15. Korea, Rep. of .......... 4.3 i’ 1.1% 16. Canada ................ 2.9 0.7% 17. NewZealand ........... 2.9 ii 0.7% 18. Brunei ................. 2.6 ,: 0.6% 19. France ................ 2.3 20. Pakistan ............... 2.1 ~" 0.5% Other ................. 23.5 Percentage of Outgoing Traffic MiTT is Minutes of Telecommunications Traffic. Data are in millions of minutes for public voice circuits. © TeleGeography, Inc. 1996 National Traffic Balance M iTT FY 1993/94 1994/95 1995/96 Incoming 304.2 399.7 442.0 Outgoing 258.1 342.3 408.3 Surplus (Deficit) 46.1 57.4 33.7 Total Volume 562.3 742.0 850.3 Notes: Traffic is for Telekorn Malaysia only. Data based on billing point of traffic. Totals may appear inconsistent with other figures due to rounding. Fiscal year ends 31 March. TeteGeography 1996/97 © TeleGeographyo Inc. 1996 orocco Destination Largest Telecommunications Routes, MiTT Percentage of Outgoing Traffic 1995 1. France ................ 57.7 2. Spain ................. 11.4 3. Italy ................... 7.1 4. Belgium ................ 6.0 5. Germany ............... 5.7 6. Netherlands ............ 5.2 7. United States ........... 4.6 8. Saudi Arabia ............ 4.2 9. United Kingdom ......... 2.8 10. Switzerland ............. 2.3 11. Canada ................ 2.3 12. Tunisia ................. 2.0 13, Algeria ................. 1.9 14. United Arab Emirates .... 1.2 15. Egypt .................. 1.0 16. Libya .................. 0.7 17. Portugal ............... 0.6 18. Denmark ............... 0.5 19. Sweden ................ 0.5 20. Senegal ................ 0.4 Other ................. 11.8 3.2% ~L 2.2% i~i 1.8% ~i!i! 1.8% I~:i 1,5% 0.9% 0.8% 0.5% 0.5% 0.4% 0.4% 0.3% MiTT is Minutes of Telecommunications Traffic. Data are in millions of minutes for public voice circuits. © Tele6eography, Inc. 1996 National Traffic Balance MiTT 1993 1994 1995 Incoming n.a. n.a. 265 Outgoing 125.1 130.0 129.7 Surplus (Deficit) n.a. n.a. 135 Total Volume n.a. n.a. 394 Note: Data based on billing point of traffic. Totals may appear inconsistent with other figures due to rounding. 140 © TeleCeography, Inc. 1996 International Telephone Traffic--Asia, Middle East, & Africa New Zealand Destination Mi’n" 1. Australia ............. 142 2. United Kingdom ........ 37 11.9% 3. United States .26 ~,~~J:~:~ 8.3% ......... ~-~:1:~:2 ~ZL 4. Hong Kong ............. 7 ~i 2.2% 5. Japan ................. 7 ~i:i 2.2% 6. Canada ................ 6 1.9% 7. Fiji .................... 5 i~" 1.s% 8. Singapore ............. 5 ~i, 1.~% 9. Malaysia .............. 5 ~i 1.8% 10. Taiwan ................ 4 ~, 1.3% 11. Western Samoa ......... 4 i~,,! 1.3% 12. Germany .............. 3 !~! 1.o% 13. China ................. 3 ~i; 1.o% 14. India .................. 3 ~.’ 1.o% 15. South Africa ............ 2 i: 0.6% Other ................. 53 il,il ~i!~i,,ilii{i :ili ;:~!:i~’~ii:, ;:i,,~: 17.0% Largest Telecommunications Routes, FY ! 995/96 Percentage of Outlining Traffic MiTT is Minutes of Telecommunications Traffic. Data are in millions of minutes for public voice circuits. © Tele6eography, Inc. 1996 National Traffic Balance MiTT FY 1993/94 1994/95 1995/96 Incoming 227 263 327 Outgoing 211 261 312 Surplus (Deficit) 16 2 15 Total Volume 438 524 639 Note: Data rounded to the nearest million minutes for Telecom New Zealand and Clear Communications Ltd. combined. Fiscal year ends 31 March. 141 TeleGeography 1996/97 © TeleGeography, Inc. 1996 Pakistan Largest Destination Mi’rr 1, United Arab Emirates ...10.6 2. United Kingdom ......... 9.4 3. United States ........... 9.1 4. Saudi Arabia ............ 6.3 5. Italy ................... 3.3 6. Japan ................. 2.4 7. Germany ............... 2.4 8. Canada ................ 1.6 9. Hong Kong ............. 1.2 10. France ................. 1.1 11. Australia ............... 0.9 12. Turkey ................. 0.8 13. Switzerland ............. 0.7 14. Korea, Rep. of ........... 0.5 15. Bahrain ................ 0.4 16. Malaysia ............... 0.3 17. Belgium ................ 0.2 Other ................. 14.9 Telecommunications Routes, 1995 Percentage of Outgoing Traffic 3.6% ~-7~; 1.4% i~7: 1.o% 0.7% 0.6% 0.4% 0.3% MiTT is Minutes of Telecommunications Traffic. Data are in millions of minutes for public voice circuits. © Tele6eography, Inc. 1996 National Traffic Balance MiTT 1993 1994 1995 Incoming 305.7 n.a. 362.1 Outgoing 56.5 61.4 65.9 Surplus (Deficit) 249.2 n.a. 296.1 Total Volume 362.2 n.a. 428.0 Note: Traffic to India and Bangladesh is excluded from route data. Totals may appear inconsistent with other fig- ures due to rounding. © TeleGeography, Inc. 1996 International Telephone Traffic--Asia, Middle East, & Africa Philippines Largest Telecommunications Routes, 1995 Destination MiTT 1. United States .......... 62 2. Japan ................ 28 3. Hong Kong ............ 16 4. Canada ............... 12 5. Australia .............. 9 6. Singapore ............. 7 7. Taiwan ................ 7 8. Korea, Rep. of .......... 5 9. Saudi Arabia ........... 5 10. United Kingdom ......... 4 11. Malaysia .............. 3 12. Italy ................... 3 Other ................. 13 Percentage of Outgoing Traffic !,~; 2.3% 1.7% 1.7% MiTT is Minutes of Telecommunications Traffic. Data are in millions of minutes for public voice circuits. © TeleGeography, Inc. 1996 National Traffic Balance MiTT 1993 1994 1995 Incoming n.a. 617 691 Outgoing 164 160 174 Surplus (Deficit) n.a. 457 517 Total Volume n.a. 777 865 Note: Data are rounded to the nearest million minutes. Totals may appear inconsistent with other figures due to rounding. 143 TeleGeography 1996/97 © TeleGeographyo Inc. 1996 Destination Largest Telecommunications Routes, 12.9 2. United Arab Emirates ...11.8 3. Egypt ................. 8.7 4. Saudi Arabia ........... 6.3 5. United Kingdom ......... 4.6 6. Bahrain ............... 4.4 7. Pakistan ............... 4.0 8. United States .2.2 ~i~: 9. Jordan .1.8 ~,~! 2.4% 10. Bangladesh .1.8 ~!~,~~’q,," ........... 2.4% 11. Kuwait ................ 1.7 ~:~2~;~?~; 2.2% 12. Iran .1.4 ~-~:’ 1.8% .................. 13. 0man .1.2 ~%~7~ ........ 1.6% ~z~;~ ~. ~,~ ................ ~.= ~:~ 1.8% 15. Philippines ............. 1.1 ~,~:~: 1.5% 16. Sudan .1.0 W;~:,:: ~.3% ................ :~,~: 17. Syria .0.9 ~’,:~:7~ 1.2% ................. ~:F~s 18. Lebanon ............... 0.9 ~:~ 1.2% 19. Italy ................... 0.8 ~L:~ ~.o% 20. Sri Lanka .0.7 ~,,~ 0.9% Other .................. 6.5 MiTT is Minutes of Telecommunications Traffic. Data are in millions of minutes for public voice circuits. © TeleGeography, Inc. 1996 National Traffic Balance MiTT 1993 1994 1995 Incoming n.a. n.a. 52.6 Outgoing 58.3 62.7 75.8 Surplus (Deficit) n.a. n.a. (23.2) Total Volume n.a. n.a. 128.4 Note: Totals may appear inconsistent with other figures due to rounding. 144 © TeleGeography, Inc. 1996 International Telephone Traffic--Asia, Middle East, & Africa F .............. 137.3 2. Pakistan .............. 56.3 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. Saudi Arabia India ................. 52.2 Syria ................. 21.3 Yemen ................ 17.1 Jordan ............... 16.2 ~:i: 3.2o/o United Kingdom ........ 15.9 3.2% United States .......... 14.3 ~!!~,,i 2.9o/o Bahrain .............. 14.1 ~:,~i 2.8% United Arab Emirates ...13.2 Philippines ............ 13.0 ~ii 2.6O/o Kuwait ............... 11.1 ~!!iI 2.2% Sudan ................ 10.9 Lebanon .............. 10.8 i’.i~i~!; 2.2% Morocco .............. 9.4 i"," 1.9% Bangladesh ............ 9.1 iii,,~. 1.8O/o Turkey ................. 8.9 iii~!,’ 1.8% France ................ 7.9 ii!’~, 1.6% Germany .............. 5.4 !:~, 1.1% Qatar ................. 3.1 i. 0.6% Other ................. 51.6 Largest Telecommunications Routes, 1995 Mil"l" Percentage of Outgoing Traffic MiTT is Minutes of Telecommunications Traffic. Data are in millions of minutes for public voice circuits. © TeleGeography, Inc. 1996 National Traffic Balance MiTT 1993 1994 1995 Incoming n.a. n.a. n.a. Outgoing 454.9 477 499.1 Surplus (Deficit) n.a. n.a. n.a. Total Volume n.a. n.a. n.a. Nnte: Totals may appear inconsistent with other figures due to rounding. 145 TeleGeography 1996/97 © TeleGeography, Inc. 1996 Singapore Destination Largest Telecommunications Routes, FY 1995/96 MiTt Percentage of Outgoing Traffic 1. Malaysia ............ 218 2. Indonesia ............. 64 3. Hong Kong ............ 62 4. United States .......... 51 5. Japan ................ 47 6. China ................. 35 7. Australia .34 ~;;~ 4.4% 8. Thailand .............. 33 10. Philippines ............ 28 3.6% 11. India ................. 28 ~~ 3.6% 12. Taiwan ............... 27 ~,,_~:~ 3.5% 13. Germany .12 ............. 14. Korea, Rep. of ......... 11 %~~ ~.4% 15. France ................ 7 16. Brunei ................. 7 ~" 0.9% Notes: MiTT is Minutes of Telecommunications Traffic. Data are in millions of minutes for public voice circuits. © TeleGeography, Inc. 1996 National Traffic Balance MiTT FY 1993/94 1994/95 1995/96 Incoming n.a. n.a. n.a. Outgoing 480 643 773 Surplus (Deficit) n.a. n.a. n.a. Total Volume n.a. n.a. n.a. Fiscal year ends 31 March. Totals may appear inconsistent with other figures due to rounding. 146 © TeleGeography, Inc. 1996 International Telephone Traffic--Asiao Middle East, & Africa South Africa Destination MiTT 1. United Kingdom ........ 40.9 2. Namibia ............... 29.9 3. Zimbabwe ............. 21.6 4. United States .......... 20.2 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. Largest Telecommunications Routes, 1994 Percentage of Outgoing Traffic Botswana ............. 14.0 Mozambique ........... 11.7 Germany .............. 11.7 Swaziland ............. 11.0 Lesotho ................ 7.9 Australia ............... 6.6 Portugal ............... 4.4 France ................. 4.4 Canada ................ 3.9 Netherlands ............ 3.8 Italy ................... 3.7 Israel .................. 3.6 Switzerland ............. 3.6 Zambia ................ 3.5 Malawi ................ 2.5 Taiwan ................. 2.4 Other ................. 51.4 MiTT is Minutes of Telecommunications Traffic. Data are in millions of minutes for public voice circuits. © TeleGeography, Inc. 1996 National Traffic Balance MiTT 1992 1993 1994 Incoming n.a. n.a. n.a. Outgoing 221.7 255.1 262.6 Surplus (Deficit) n.a. n.a. n.a. Total Volume n.a n.a. n.a. Note: Totals may appear inconsistent with other figures due to rounding. 147 TeleGeography 1996/97 © TeleGeography, Inc. 1996 Lanka Destination Largest Telecommunications Routes, 1. India .................. 4.0 2, United Kingdom ......... 3.0 3. Singapore ............. 2.3 4. United States ........... 1.8 5. Japan ................. 1.7 6. Hong Kong ............. 1.5 7. Australia .............. 1.2 8. Germany .............. 1.2 9. Korea, Rep. of .......... 1.0 10, United Arab Emirates .... 0,9 11. Italy ................... 0.7 12. France ................ 0.6 13. Saudi Arabia ........... 0.6 14. Canada ................ 0.6 15. Pakistan ............... 0.5 16. Maldives ............... 0.5 17. Kuwait ................ 0.5 18. Malaysia .............. 0.5 19. Thailand ............... 0.5 20. Switzerland ............. 0.4 Other ................. 3.5 1995 Percentage of Outgoing Traffic 1.8% 1.8o/o ..... 1.8% 1.5% MiTT is Minutes of Telecommunications Traffic. Data are in millions of minutes for public voice circuits. © Tele6eography, Inc. 1996 National Traffic Balance MiTT 1993 1994 1995 Incoming 65.0 78.7 92.0 Outgoing 19.5 23.7 27.5 Surplus (Deficit) 45.5 55.0 64.5 Total Volume 84.5 102.4 119.5 Note: Data based on billing point of traffic. Totals may appear inconsistent with other figures due to rounding. © TeleGeography, Inc. 1996 International Telephone Traffic--Asia, Middle East, & Africa Syria Largest Destination Mi’n" 1. Lebanon ............... 4.7 2. Jordan ................ 4.6 3. Kuwait ................ 4.1 4. United States ........... 2.5 5. Russia ................. 2.3 6. France ................ 2.3 7. Egypt ................. 2.3 8. Germany .............. 1.8 9. Saudi Arabia ........... 1.7 10. United Kingdom ......... 1.5 Other ................. 12.2 Telecommunications Routes, 1994 Percentage of Outgoing Traffic ~?~:~5~:~; 6.3% ~?~.~ 5.7% 3.8% Mil-[ is Minutes of Telecommunications Traffic. Data are in millions of minutes for public voice circuits. © Tele6eography, Inc. 1996 National Traffic Balance MiTT 1993 1994 1995 Incoming 59.3 78 n.a. Outgoing 36.7 40 66 Surplus (Deficit) 22.6 38 n.a. Total Volume 96.0 118 n.a. Note: Totals may appear inconsistent with other figures due to rounding. Route data for 1995 unavailable as of this printing. 149 TeleGeography 1996/97 © TeleGeographyo Inc. 1996 Destination China ............... 140.9 United States ......... 1 t 1.0 Hong Kong ............ 75.5 Japan ................ 66.4 Thailand .............. 22.2 Largest Telecommunications Routes, FY 1995/96 MiTr Percentage of Outgoing Traffic Philippines ............ 20.3 Canada ............... 20.2 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. Singapore ............ 17.6 Malaysia ............. 11.7 Indonesia ............. 11.6 Australia .............. 11.3 Vietnam ............... 9.0 ~,~:; 1.5% United Kingdom ......... 8.1 ~:~--1.4% Germany .............. 8.0 ~i~J 1.3% Korea, Rep. of .......... 7.7 i~ii~’! 1.3% France ................ 4.7 0.8% New Zealand ........... 4.3 !ii~;:j o.7% Italy ................... 2.9 ,12 o.5% Macau ................. 2.6 !. 0.4% South Africa ............ 2.5 0.4% Other ................. 34.4 i~’;,~i~!’~ 5.8% MiTT is Minutes of Telecommunications Traffic. Data are in millions of minutes for public voice circuits. © Tele6eography, Inc. 1996 National Traffic Balance MiTT FY 1993/94 FY 1994/95 FY 1995/96 Incoming 490.8 613.5 545.3 Outgoing 440.7 498.5 592.8 Surplus (Deficit) 50.1 115.0 (47.5) Total Volume 931.5 1,112.0 1,138.1 Note: Data based on billing point of traffic. Fiscal year ends March 31. Totals may appear inconsistent with other figures due to rounding. 150 © TeleGeography, Inc. 1996 International Telephone Traffic--Asia, Middle East, & Africa Thailand Destination Largest Telecommunications Routes, 1. United States .......... 36.0 2. Japan ................ 35.5 3. Singapore ............ 22.8 4. Hong Kong ............ 16.3 5. Taiwan ............... 13.9 6. United Kingdom ........ 11.7 7. Australia .............. 9.3 8. China ................. 8.1 9. Germany .............. 7.9 10. Korea, Rep. of .......... 6.4 11. France ................ 4.7 12. India .................. 4.5 13. Italy ................... 3.6 14. Burma ................. 3.3 15. Indonesia .............. 3.1 16. Philippines ............. 3.0 17. Switzerland ............ 2.8 18. Canada ................ 2.7 19. Vietnam ............... 2.4 20. Netherlands ............ 2.3 Other ................. 18.4 1995 Percentage of Outgoing Traffic MiTT is Minutes of Telecommunications Traffic. 1.2% Data are in millions of minutes for public voice circuits. © Tele6eography, Inc. 1996 National Traffic Balance MiTT 1993 1994 1995 In coming 218.7 313.3 277.7 0 utgoing 161.8 173.2 218.8 Surplus (Deficit) 56.9 140.1 58.9 Total Volume 380.5 486.5 496.5 Note: Totals may appear inconsistent with other figures due to rounding. 151 TeleGeography 1996/97 © TeleGeography, Inc. 1996 United Arab Emirates Destination Largest Telecommunications Routes, 1995 Mi~l" Percentage of Outgoing Traffic India ,108.4 ~,-’~,~v~-,~,,~,~,~,’, ,~ ~,~,,~ -,~,,~ ...... ""~’:~ 21.5% Pakistan .49.0 ........... .. ;~:~ ~:~;~.~ ~: 9.7% United Kingdom ........ 33.5 Egypt .33.2 "~T~"~~":T~ 6.6% ................. Saudi Arabia .30.9 .......... ~-/~ -:,~ :~’,~:~:~:~ 6.1% 4.6% ~:~ 4.2% ,~: 2.6% ~: 1.5% ~:~ 1.3% -1.2% 1.2% 1 .o% Other ................. 70.2 Mi~ is Minutes of Telecommunications Traffic. Bata are in millions of minutes for public voice circuits. @ TeleGeography, Inc. 1996 2. 3. 4. 5. 6. United States .......... 23.4 7. Oman ................. 21.0 8. Syria ................. 18.7 9. Iran .................. 16.7 10. Qatar ................. 13.1 11. Jordan ................ 12.7 12. Bahrain ............... 12.1 13. Kuwait ................ 11.3 14. Philippines ............. 9.4 15. Lebanon ............... 9.0 16. Bangladesh ............. 7.8 17. Sudan ................. &3 18. France ................. 6.1 19. Germany ............... 5.9 20. Yemen ................. 4.9 National Traffic Balance MiTT 1993 1994 1995 Incoming n.a. n.a, n.a. Outgoing 359.0 428.2 503.6 Surplus (Deficit) n.a. n.a. n.a. Total Volume n.a. n.a. n.a. 152 © TeleGeagraphy, Inc. 1996 International Telephone Traffic--Asia, Middle East. & Africa Vietnam Largest Destination MiTT 1. China .................. 6.0 2. Taiwan ................. 3.0 3. United States ........... 3.0 4. Philippines ............. 2.0 5. Hong Kong ............. 2.0 6. France ................. 2.0 7. Korea, Rep. of ........... 1.5 8. Singapore .............. 1.5 9. Thailand ............... 1.5 10. Germany ............... 1.0 Other ................. 11.5 Telecommunications Routes, Percentage of Outgoing Traffic 17.1% 1995 MiTT is Minutes of Telecommunications Traffic. Data are in millions of minutes for public voice circuits. © Tele6eography, Inc. 1996 National Traffic Balance MiTT 1993 1994 1995 Incoming n.a. n.a. 170.3 Outgoing 14 24 35.1 Surplus (Deficit) n.a. n.a. 135.2 Total Volume n.a, n.a. 205.4 Note: Data rounded to the nearest million minutes. 153 TeleGeography 1996/97 © TeleGeography, Inc. 1996 Methodology and Sources T he traffic statistics in TeleGeography 1996/97 were compiled primarily from an independent survey of telecommuni- cations service providers by TeleGeography, Inc. (TGII. For some countries and carriers, traffic data have been estimated based upon annual reports, gov- ernment publications and industry interviews. See the footnotes to each table for further infor- mation. Direction of Traffic 1996, jointly com- piled by TGI and the International Telecommunication Union (ITU), was also consult- ed. Traffic volumes in TeleGeography 1996/97 are generally reported in minutes or Mir-l (Minutes of Telecommunications Traffic). In most cases MiTT refer to paid minutes of traffic on public switched voice circuits and thus includes voice as well as non-voice (facsimile, data) traffic. For the origins of MiTT and its various applications (economic forecasting, competition policy, geographyl, see G. Staple and M. Mullins, "Telecom Traffic Statistics--MiTT Matter," Telecommunications Policy, Vol. 14, No. 2, June 1989; and G. Staple "The new demand for telecoms traffic data: from MiTT to maps," Telecommunications Policy, Vol. 20, No. 9, October 1996. Calling Card Traffic Historically, most international calls were billed at the point of origination. The number of billed minutes thus coincided with the volume of outgo- ing traffic. Billed minutes also included collect or reverse charge calls because the calls were set up by an operator in the originating country. However, the recent use of credit and debit cards has shifted the billing point for many internation- al calls. For example, calls from Italy to the United States (or a third country, such as Argentina) may now be set up and billed in the U.S. Unless otherwise stated in the notes to a table, the outbound MiTT reported for countries in TeleGeography 1996/97 refers to outbound traf- fic originated in the reporting country even if it is billed in another country. Some countries, including the U.S., report inter- national traffic data based solely on the location where the traffic is billed. Consequently, "out- bound" traffic data for these countries includes Home Country direct traffic originating in third countries (e.gl, a call originated in Italy to a U.S. number and billed to a U.S. calling cardl. For these and other reasons (such as different fiscal years), the national statistics in TeleGeography are not directly comparable, and incoming MiTT reported for one country may not match the out- going MiTT on the same route by the correspon- dent country. Some double counting may also occur. For example, a Country Direct call from Spain to the U.S. which is billed to a U.S. calling card is reported here as outbound U.S. Mil-~ the same call also is reported as outbound MiTT by Spain. Third-Country Routing The growing volume of traffic routed via a third country using Home Country Beyond and "call back" services is also making national traffic sta- tistics harder to interpret. A Home Country Beyond call may originate in Country A, be billed to a calling card in Country B and terminate in Country C. Similarly, a call routed via a "call back" service may be placed by a subscriber in Country A, but originate in Country B and termi- nate in Country C. In both cases, the calls from Country A to Country C generally will not be counted in Country A~s outbound MiTT but may be reflected by an increased volume of MiTT from Country B to Country C. Accordingly, in countries where Home Country Beyond and call back services are widely used, a year-to-year comparison of national MiTT also requires examining the statistics of countries, such as the U.S., where the calls are being refiled or hubbed. 154 © TeleGeography, Inc. 1996 Internattonal Telephone Traffic--Methodology and Sources To assist readers in making such comparisons, the U.S. tables in TeleGeography 1996/97 have been expanded to provide 1994 and 1995 route-by- route statistics for over sixty countries. Resellers The MiTT data in this report includes traffic car- ried by resellers of international switched voice services. Such traffic is counted as part of the MiTT for the facilities based carrier whose facilities are resold. Traffic carried by International Simple Resale (ISR] carriers is excluded, unless otherwise stated. ISR carriers resell the capacity of international private lines (IPLsl for switched services by interconnect- ing their IPLs to the public switched network at one or both ends. The report also excludes transit traffic--that is, traffic which merely passes through a given coun- try, but is not refiled via the switched network in the reporting country. Other Factors traffic data for the originating country. For exam- ple, neighboring countries may not classify local cross border traffic in the same way (i.e., one country may count all such traffic as internation- al, while the other does not.) In any event, the route-by-route traffic data reported in TeleGeography for each country generally is based upon the survey data supplied to TeleGeography by the originating country, not the terminating country. Some differences exist between the historical sta- tistics (1994 or earlier) reported in TeleGeography 1996/97 and data stated in prior reports or Direction of Traffic. The variations reflect corrections and/or revised data subse- quently provided to TeleGeography. tn addition, rounding may cause the figures on total national traffic and surpluses and deficits to appear incon- sistent with other national data. ¯ There may also be other reasons (beyond those referred to above) which cause inbound traffic data on a given route to differ from the outbound 155 TeleGeography 1996/97 © TeleGeographyo Inc. 1996 156 Blue Pages © TeleGeographg, Inc. 1996 Blue Pages International Dialing Codes, by Number 1 Canada United States 1-242 Bahamas 1-240 Barbados 1-268 Antigua & Barbuda 1-345 Cayman Islands 1-441 Bermuda 1-684 Montserrat 1-758 St. Lucia 1-787 Puerto Rico 1-809 Anguilla, British Virgin Islands, Dominic& Dominican Republic, Grenada, Jamaica, St. Vincent & the Grenadines, Trinidad & Tobago, Turks & Caicos, U.S. Virgin Islands 1-869 St. Kitts & Nevis islands 20 Egypt 212 Morocco 213 Algeria 216 Tunisia 218 Libya 220 Gambia 221 Senegal 222 Mauritania 223 Mali 224 Guinea 225 Ivory Coast 226 Burkina Faso 227 Niger 228 Togo Benin Mauritius 23’1 Liberia Sierra Leone 233 Ghana 234 Nigeria 235 Chad 238 Central African Republic 237 Cameroon 238 Cape Verde Islands 239 SaoTome and Principe 240 Equatorial Guinea 241 Gabon 242 Congo 243 Zaire 244 Angola 245 Guinea-Bissau 246 Diego Garcia 24? Ascension Island 248 Seychelles 249 Sudan 250 Rwanda 251 Ethiopia 252 Somalia 253 Djibouti 254 Kenya 255 Tanzania 256 Uganda 257 Burundi 258 Mozambique 259 Zanzibar 260 Zambia 261 Madagascar 262 Reunion Island 263 Zimbabwe 264 Namibia 265 Malawi 266 Lesotho 267 Botswana 268 Swaziland 269 Comoros & Mayotte 27 South Africa 290 St. Helena 291 Eritrea 297 Aruba 298 Faroe Islands 299 Greenland 30 Greece :31 Netherlands :32 Belgium 33 France 33-9:3 Monaco :34 Spain :350 Gibraltar :351 Portugal; Azores :352 Luxembourg :353 Ireland :354 Iceland 335 Albania 356 Malta 357 Cyprus 350 Finland 359 Bulgaria 36 Hungary 370 Lithuania 371 Latvia 372 Estonia 373 Moldova 374 Armenia 375 Belarus 376 Andorra 377 Monaco (reserved 378 San Marino 379 Vatican City 380 Ukraine 381 Yugoslavia 385 Croatia 386 Slovenia 387 Bosnia-Hercegovlna 389 Macedonia 39 Italy 40 Romania 41 Switzerland 41-75 Liechtenstein 42 Czech Republic 42 Slovak Republic 43 Austria 44 United Kingdom 45 Denmark 45 Sweden 47 Norway 40 Poland 45 Germany 500 Falkland Islands 501 Belize 502 Guatemala 503 El Salvador 504 Honduras 505 Nicaragua 506 Costa Rica 507 Panama 508 St. Pierre & Miquelon 509 Haiti 51 Peru 52 Mexico 53 Cuba ,54 Argentina 55 Brazil 56 Chile 57 Colombia 58 Venezuela 550 Guadeloupe 591 Bolivia 592 Guyana 553 Ecuador 594 French Guiana 595 Paraguay 596 Martinique 597 Suriname 598 Uruguay 599 Netherlands Antilles 68 Malaysia 61 Australia 62 indonesia 53 Philippines 64 New Zealand 65 Singapore 66 Thailand 670 Northern Marianas 671 Guam 672 Australian Territories 673 Brunei 674 Nauru 675 Papua New Guinea 676 Tonga Islands 877 Solomon Islands 678 Vanuatu 679 Fiji Paiau 981 Wallis & Futuna 682 Cook Islands 683 Niue 684 American Samoa 685 Western Samoa 686 Kiribati 687 New Caledonia 688 Tuvalu 689 French Polynesia 690 Tokelau 691 Micronesia 692 Marshall Islands 7 Kazakhstan Kyrgyzstan Russia Tajikistan Turkmenistan Uzbekistan 81 Japan 82 South Korea 84 Vietnam 850 North Korea 852 Hong Kong 853 Macau 855 Cambodia 1150 Laos 116 China 871 Inmarsat EastAttantic 872 Inmarsat Pacific 1173 Inmarsat Indian 874 InmarsatWest Atlantic 880 Bangladesh 886 Taiwan 90 Turkey 91 India 92 Pakistan 53 Afghanistan 94 Sri Lanka 95 Myanmar (Burma) 960 Maldives 961 Lebanon 962 Jordan 962 Syria 964 Iraq 965 Kuwait 968 Saudi Arabia 967 Yemen 968 Oman 971 United Arab Emirates 972 Israel 973 Bahrain 974 Qatar 975 Bhutan 976 Mongolia 577 Nepal 98 Iran 994 Azerbaijan 995 Georgia TeleGeography 1996/97 © TeleGeogrophU, Inc. 1996 International Dialing Codes, by Country Afghanistan ........... 93 Albania .............. 355 ~rana ............ 42 Algeria .............. 213 Algiers ............ 2 American Samoa ...... 684 Andorra ............. 376 Angola .............. 244 Luanda ............ 2 Anguilla ........... 1-809 Antigua & Barbuda ,1-268 Argentina ............. 54 Buenos Aires ........ 1 Armenia ............. 374 Yerevan ......... 8852 Aruba ............... 297 Ascension Island ..... 247 Australia ............. 61 Canberra .......... 62 Melbourne .......... 3 Sydney ............. 2 Australian Territories .672 Austria ............... 43 Vienna ............. 1 Azerbaijan ........... 994 Baku ........... 8922 Bahamas ........... 1-889 Bahrain .............. 973 Bangladesh .......... 880 Dhaka ............. 2 Barbados ........... 1-809 Belarus .............. 375 Minsk ........... 172 Bel0ium .............. 32 Brussels ........... 2 Belize ............... 581 Belmopan .......... 8 Benin ............... 229 Bermuda ........... 1-441 Bhutan .............. 975 Bolivia .............. 551 La Paz ............. 2 Bosnia .............. 387 Sarajevo ........... 71 Botswana ............ 267 Brazil ................ 55 Brasilia ........... 81 Rio de Janeiro ...... 21 S~o Paulo .......... 11 British Virgin Islands 1-809 Brunei ............... 673 Bandar Seri Begawan..2 Bulgaria ............. 359 Sofia .............. 2 Burkina Faso ......... 226 Burundi .............. 257 Cambodia ............ 855 Cameroon ............ 237 Canada ................ 1 Montreal ......... 514 Ottawa ........... 613 Toronto ........... 416 Cape Verde .......... 238 Cayman Islands ..... 1-345 Central African Republic ............. 236 Bangui ............ 61 Chad ................ 235 Chile ................. 56 Santiago ........... 2 China, People’s Republic of ........ 86 Beijing ............. Guangzhou ......... 20 Shanghai .......... 21 Colamhia ............. 57 Bogota ............ 1 Cocos Islands; Norfolk & Christmas Islands .672 Comoros ............. 269 Congo ............... 242 Brazzaville .... 81/82/83 Costa Rica ........... 506 Croatia .............. 385 Zagreb ............. 1 Coha ................. 53 Havana ............ 7 Cyprus ............... 357 Nicosia ............ 2 Czech Republic ........ 42 Prague ............. 2 Benmark .............. 45 Diego Garcia ......... 246 Djibouti .............. 253 Dominica ........... 1-809 Dominican Republic .1-809 Ecuador ............. 593 Quito .............. 2 Egypt ................. 20 Cairo .............. 2 El Salvador ........... 503 Equitorial Guinea ..... 240 Eritrea ............... 291 Estonia .............. 372 Tallinn ............. 2 Ethiopia ............. 251 Addis Ababa ........ 1 Falkland Islands ...... 500 Faroe Islands ......... 298 Fiji ................. 679 Finland .............. 358 Helsinki ............ 9 France ................ 33 Paris .............. 1 French Antilles ....... 596 French Guiana ........ 594 French Polynesia ..... 689 Gabon ............... 241 Gambia .............. 220 Georgia .............. 995 Tbilisi .......... 8832 Germany .............. 49 Berlin ............. 30 Bonn ............ 228 Frankfurt ........... 69 Munich ............ 89 Ghana ............... 233 Accra ............. 2! Gibraltar ............. 350 Greece ............... 30 Athens ............ 1 Greenlaod ........... 299 Grenada ........... 1-809 6uadelaape .......... 590 G,,am ................ 671 6oatemala ........... 502 Guatemala City ...... 2 Gainea .............. 224 Guinea-gissau ........ 245 Guyana .............. 592 Georgetown ........ 2 Haiti ................ 509 Honduras ............ 504 Hong Kong ........... 852 Hungary .............. 36 Budapest ........... 1 Iceland .............. 354 India ................. 91 Bombay ........... 22 Calcutta ........... 33 New Delhi ......... 11 Indonesia ............ 62 Jakarta ........... 21 Inmarsat East Atlantic ....... 871 West Atlantic ...... 874 Pacific ........... 872 Indian ............ 873 Iran .................. 98 Tehran ............ 21 Iraq ................. 964 Baghdad ........... 1 Ireland .............. 353 Dublin ............. 1 Israel ................ 972 Jerusalem .......... 2 Tel Aviv ............ 3 Italy .................. 39 Rome .............. 6 Milan .............. 2 Ivory Coast ........... 225 Jamaica ........... 1-809 Japan ................ 81 Osaka .............. 6 Tokyo .............. 3 Jordan .............. 962 Amman ............ 6 Kazakhstan ............ 7 Alma Ata ........ 3272 Kenya ............... 254 Nairobi ............ 2 Kiribati .............. 686 Kuwait .............. 965 Kyrgyzstan ............. 7 Bishkek ......... 3312 Laos ................ 856 Latvia ............... 371 Riga ............... 2 Lebanon ............. 981 Beirut ............. 1 Lesotho .............. 266 Liberia .............. 231 Libya ................ 218 Tripoli ............ 21 Liechtenstein ....... 41-75 Lithuania ............ 378 Vilnius ............. 2 Luxembourg .......... 352 Macau .............. 853 Macedonia ........... 389 Skopje ............ 91 Madagascar .......... 261 Antananarivo ........ 2 Malawi .............. 265 Malaysia ............. 60 Kuala Lumpur ....... 3 Maldi~es ............ 960 Mali ................ 2.23 Malta ............... 356 Martinique ........... 596 Mauritania ........... 222 © TeleGeography, Inc. 1996 Blue Pages Mauritius ............ 230 Mayotte ............. 269 Mexico ............... 52 Guadalajara ........ 36 Mexico City ......... 5 Monterrey ......... 83 Micronosia ........... 691 Moldova ............. 373 Chisinau .......... 422 Monaco ............ 33-93 Mongolia ............ 976 Montserrat ......... 1-664 Morocco ............. 212 Casablanca .......... 2 Babat ............. 7 Mozambique ......... 258 Maputo ............ 1 Myanmar (Burma) ..... 95 Namibia ............. 264 Windhoek ......... 61 Nauru ............... 674 Nepal ............... 977 Kathmandu .......... 1 Netherlands ........... 31 Amsterdam ........ 20 Netherlands Antilles .599 Nevis Islands ....... 1-069 New Caledonia ....... 607 New Zealand .......... 64 Auckland ........... 9 Wellington ......... 4 Nicaragua ........... 505 Managua ........... 2 Nigor ................ 22? Ni0eria .............. 234 Lagos .............. 1 Niue ................ 603 North Korea .......... 850 Pyongyang .......... 2 Northern Marianas .... 670 Saipan ........... 322 Norway ............... 47 Oslo ............... 2 Oman ................ 968 Pakistan .............. 92 Islamabad ......... 51 Palau ................ 680 Panama ............. 507 Papua New Guinea .... 675 Paraguay ............ 595 Asuncion .......... 21 Peru ................. 51 Lima ............. 14 Philippines ............ 63 Manila ............ 2 Poland ............... 48 Warsaw .......... 22 Portugal ............. 351 Lisbon ............. 1 Puerto Rico ......... 1-809 Qatar ................ 974 Reunion Island ....... 262 Romania .............. 40 Bucharest .......... 1 Russia ................. 7 Moscow ......... 095 St. Petersburg ...... 812 Rwanda ............. 250 St. Kilts ............ 1-869 St. Lucia ........... 1-758 St. Pierre & Miquelon..508 St. Vincent & ........... the Grenadines..1-809 San Marino .......... 378 S~o Tome and Principe 239 Saudi Arabia ......... 966 Riyadh ............. 1 Senegal ............. 221 Seychelles ........... 248 Sierra Leone ......... 232 Freetown .......... 22 Singapore ............. 65 Slovak Republic ....... 42 Bratislava .......... 7 Slovenia ............. 386 Ljubljana .......... 61 Solomon Islands ...... 677 Somalia ............. 252 Mogadishu ......... 1 South Africa ........... 27 Johannesburg ....... 11 Pretoria ........... 12 South Korea ........... 82 Seoul .............. 2 Spain ................ 34 Madrid ............ 1 Barcelona ........... 3 Sri Lanka ............. 94 Colombo ........... 1 Sudan ............... 249 Khartoum .......... 11 Suriname ............ 597 Swaziland ........... 268 Sweden .............. 46 Stockholm .......... 8 Switzerland ........... 41 Berne ............. 31 Zurich .............. 1 Syria ................ 963 Damascus ......... 11 Tahiti ................ 689 Taiwan .............. 886 Taipei ............. 2 Tajikistan .............. 7 Dushanbe ....... 3772 Tanzania ............. 255 Dar Es Salaam ...... 51 Thailand .............. 66 Bangkok ........... 2 Togo ................ 228 Tokelau .............. 690 Tonga ............... 676 Trinidad & Tobago ...1-809 Tunisia .............. 216 Tunis .............. 1 Turkey ................ 90 Ankara ............ 4 Istanbul ............ 1 Turkmenistan ........... 7 Ashkhabad ....... 3632 Turks & Caicos ..... 1-809 Tuvalu ............... 688 Uganda .............. 256 Kampala .......... 41 Ukraine .............. 380 Kiev .............. 44 United Arab Emirates .971 Abu Dhabi .......... 2 Dubai .............. 4 United Kingdom ........ 44 London ....... 171/181 Manchester ....... 161 United States ........... 1 Chicago ....... 312/630 Houston .......... 713 Los Angeles ....... 213 Miami ............ 305 New York ..... 212/718 Washington ....... 202 Uruguay ............. 598 Montevideo ......... 2 Uzbekistan ............. 7 Tashkent ........ 3712 Vanuatu ............. 678 Vatican Ci~ .......... 379 Venezuela ............ 58 Caracas ............ 2 Vietnam .............. 84 Wallis & Futuna ....... 681 Western Samoa ....... 685 Yemen ............... 967 Sanaa ............ 51 Yugoslavia ........... 381 Belgrade ........... 11 Zaire ................ 243 Kinshasa .......... 12 Zambia .............. 260 Lusaka ............. 1 Zanzibar (Tanzania) .... 259 Zimbabwe ........... 263 Harare ............. 4 TeleGeography1996/97 © TeleGeograph~l, Inc. 1996 North American Area Codes, by State and Province Alabama Birmingham ....... 205 Montgomery ...... 334 Alaska .............. 907 Alberta ............. 403 Arizona Phoenix .......... 602 Tucson ........... 520 Arkansas ............ 501 Bahamas ............ 242 Barbados ........... 246 Bermuda ............ 441 British Columbia Victoria .......... 250 Vancouver ........ 604 California Anaheim ......... 714 Bakersfield ....... 805 Burbank ...... 818/562~ Fresno ........... 209 Long Beach ...310/5621 Los Angeles...213/562~ Oakland .......... 510 Riverside ......... 909 Sacramento ....... 916 San Diego .... 619/760~ San Francisco ..... 415 San Jose ......... 408 Santa Rosa ....... 707 Caribbean ........... 809 Colorado Colorado Springs ...719 Denver ........... 303 Ft. Collins ........ 970 Connecticut Bridgeport ........ 203 Hartford .......... 860 Delaware ........... 302 District of Columbia Washington ....... 202 Florida Ft. Myers ......... 941 Gainesville ........ 352 Jacksonville ....... 904 Miami ........ 305/954 Orlando .......... 407 Tampa ........... 813 Georgia Athens ........... 706 Atlanta .......... 404 Marietta ......... 770 Savannah ......... 912 Hawaii ............. 808 Idaho ............... 208 Illinois Aurora ........... 630 Cairo ............ 618 Chicago .......... 312 Chicago .......... 773 Evanston ......... 847 Oak Brook ........ 708 Peoria ........... 309 Rockford ......... 815 Springfield ........ 217 Indiana Evansville ........ 812 Gary ............. 219 Indianapolis...317/7652 Iowa Council Bluffs ..... 712 Des Moines ....... 515 Dubuque ......... 319 Kansas Topeka ........... 913 Wichita .......... 316 Kentucky Dade Park ........ 812 Lexington ......... 606 Louisville ......... 502 Louisiana New Orleans ...... 504 Shreveport ........ 318 Maine .............. 207 Manitoba ........... 204 Map/land Baltimore ......... 410 Rockville ......... 301 Massachusetts Boston ........... 617 Springfield ........ 413 Worcester ........ 508 Michigan Detroit ........... 313 Flint ............. 810 Grand Rapids ...... 616 Lansing .......... 517 Sault Ste. Marie ...906 Minnesota Duluth ........... 218 Minneapolis ....... 612 Rochester ........ 507 St. Cloud ......... 320 Mississippi .......... 601 Missouri Jefferson ......... 573 Kansas City ....... 816 St. Louis ......... 314 Springfield ........ 417 Montana ............ 406 Nebraska North Platte ....... 308 Omaha ........... 402 Nevada ............. 702 New Brunswick ...... 506 New Hampshire ...... 603 New Jersey Elizabeth ......... 908 Newark .......... 201 Trenton .......... 609 New Mexico ........ 505 New York Albany ........... 518 Bronx, Queens .718/917 Buffalo ........... 716 Hempstead ....... 516 Ithaca ........... 607 Manhattan .... 212/917 Syracuse ......... 315 White Plains ...... 914 Newfoundland ....... 709 North Carolina Charlotte ......... 704 Greensboro ....... 910 Raleigh .......... 919 North Dakota ........ 701 Nova Scotia & Prince Edward Island .... 902 Ohio Cincinnati ..... 513/937 Cleveland ......... 216 Columbus ......... 614 Toledo ........... 419 Oklahoma Oklahoma City ..... 405 Tulsa ............ 918 Ontario Ft. William ....... 807 London ........... 519 North Bay ........ 705 Ottawa .......... 613 Toronto .......... 416 Oregon Eugene .......... 541 Portland .......... 503 Pennsylvania Altoona .......... 814 Harrisburg ........ 717 Philadelphia ....... 215 Pittsburgh ........ 412 Puerto Rico .......... 787 Ouebec Montreal ......... 514 Quebec .......... 418 Sherbrooke ....... 819 Bhode Island ........ 401 Saskatchewan ....... 306 South Carolina Charleston ........ 803 Greenville ........ 864 South Dakota ........ 605 Memphis ......... 901 Nashville ......... 615 Knoxville ......... 423 Texas Amarillo .......... 806 Austin ........... 512 Dallas ........ 214/972 El Paso .......... 915 Fort Worth ........ 817 Galveston ........ 409 Houston ...... 713/281 San Antonio ....... 210 Tyler ............. 903 Utah ................ 801 Vermont ............. 802 Virginia Alexandria ........ 703 Richmond ......... 804 Roanoke ......... 540 Norfolk .......... 757 Washington Olympia .......... 360 Seattle ........... 206 Spokane ......... 509 West Virginia ........ 304 Wisconsin Madison ......... 608 Milwaukee ....... 414 Eau Claire ........ 715 Wyomin0 ............ 307 Two codes separated by a slash (e.g., in Dallas, Texas)indicate an overlay; multiple codes are used for the same geographic area. 1 25 January 1997 2 1 February 1997 3 22 March 1997 © TeleGeography, Inc. 1996 Blue Pages North American Area Codes, by Number 21)1 New Jersey 401 Rhode lsland 61)2 Arizona 21)2 District of 402 Nebraska 61)3 New Hampshire Columbia 403 Alberta 61)4 British Columbia 203 Connecticut 404 Georgia 605 South Dakota 204 Manitoba 405 Oklahoma 606 Kentucky 205 Alabama 406 Montana 607 New York 206 Washington 407 Florida 608 Wisconsin 207 Maine 408 California 609 New Jersey 208 Idaho 409 Texas 610 Pennsylvania 209 California 410 Maryland 612 Minnesota 210 Texas 412 Pennsylvania 613 Ontario 212 New YorkCity 413 Massachusetts 614 Ohio 213 California 414 Wisconsin 615 Tennessee 214 Texas 415 California 616 Michigan 215 Pennsylvania 416 Ontario 617 Massachusetts 216 Ohio 417 Missouri 618 Illinois 217 Illinois 418 Quebec 619 California 218 Minnesota 419 Ohio 630 Illinois 219 Indiana 423 Tennessee 664 Montserrat 242 Bahamas 441 Bermuda 701 North Dakota 246 Barbados 500 Personal Communication 702 Nevada 250 British Columbia Services(PCS) 703 Virginia 268 Antigua & Barbuda 501 Arkansas 704 North Carolina 281 Texas 502 Kentucky 705 Ontario 301 Maryland 503 Oregon 706 Georgia 31)2 Delaware 504 Louisiana 707 California 303 Colorado 505 New Mexico 708 Illinois 304 West Virginia 506 New Brunswick 709 Newfoundland 305 Florida 507 Minnesota 710 U.S. GovernmentErnergency 306 Saskatchewan 508 Massachusetts Telecomrnunications Service 307 Wyoming 509 Washington 712 Iowa 301) Nebraska 510 California 713 Texas 309 Illinois 512 Texas 714 California 310 California 513 Ohio 715 Wisconsin 312 Illinois 514 Quebec 716 New York 313 Michigan 515 Iowa 717 Pennsylvania 314 Missouri 516 New York 718 New York City 315 New York 517 Michigan 719 Colorado 316 Kansas 518 New York 757 Virginia 317 Indiana 519 Ontario 751) St. Lucia 318 Louisiana 520 Arizona 760 San Diego (3 March 97) 319 Iowa 540 Virginia 765 Indiana (1 Feb. 97) 320 Minnesota 541 Oregon 770 Georgia 331) Ohio 555 Public Information Services 773 Illinois 334 Alabama 561 Florida 71)7 Puerto Rico 345 Cayman Islands 562 California(25Jan.97) 800 Toll-freeservices 352 Florida 573 Missouri 801 Utah 360 Washington 601 Mississippi 802 Vermont 803 South Carolina 804 Virginia 805 California 806 Texas 807 Ontario BOB Hawaii 809 Puerto Rico and Caribbean 810 Michigan 812 Indiana/Kentucky 813 Florida 814 Pennsylvania 815 Illinois 816 Missouri 817 Texas 818 California 819 Quebec 847 Illinois 860 Connecticut 864 South Carolina 869 St. Kitts & Nevis 81)0 Toll-free services 881 Toll-free services 881) Toll-free services 500 Information Services 901 Tennessee 902 Nova Scotia and Prince Edward Island 903 Texas 904 Florida 905 Ontario 906 Michigan 907 Alaska 908 New Jersey 909 California 910 North Carolina 912 Georgia 913 Kansas 914 New York 915 Texas 916 California 917 New York City 918 Oklahoma 919 North Carolina 537 Cincinnati 941 Florida 970 Colorado 972 Texas TeleGeo~raphy1996/97 © TeleGeogmphg, Inc. 1996 N~ADA ~2 UTAH Phoenix Wlnr pete 219 Minneapo~ To(onto N ORll4 DAKOTA 328 $1~27t5 SOUTH DAKOTA WISG~SIN mWA 814 ~2 913 ~4~ 618 ;INLAND RUSSIA MEXICO CARIBBF.AN NaSsau, |EXCEI~ATED| F242 Ah’TIGUA & BARBUDA 5) B~ PUERTO RIC0 ~NIQUE ELSALV~OR ~; N~6UA ARU~ ~7 ~BADOS PA~ ~NEUE~. SURINAME~7 ¯ ~.~02] PERU ~,(14] BOLIVIA La Paz (21 591 CHJLE: WESTERN SAJtARA MAURITANIA 6AMBIA ~ar BUnK LIB~:.,~ ~MARSAT MOBILE SERVICES WEST A’rZANT~C OCEAN EAST ATLANTIC OCEAN BRAZIL Rio de Janeir~ PARAGUAY S)o Paul0 (11) o ARGENTINA r~ontev,dim(2) Aires [11 © TeleGeography, Inc. 1996 Blue Pages World Telephone Codes © TeleGeography, Inc, 1996 2/6 LIBYA NIGER ~A ~’~, ".-Lagos[1) REPUBUC ETHIOPIA KENYA CONGO 2,~,~ ° ~4 INMARSAT MOBILE SERVICES COMOROS ANGOLA 244 ZIMBABWE BOTSWANA MADAGASCAR 267 261 Windhoet Gaborone MOZAMBIQUE ,112) ~,~ Maoum ’LESOTHO KAZAKHSTAN 7 MONGOUA ¯ ulaanoaa~ar 3272 GEORG~ gZBEKISTAN .... ~, TURKEY 2 ~12] ,7 Dushanbe 7 Cair~ JORDAN KUWAIT YEMEN ~mba~iZ2J ¯ 1,J NORTH KOREA Shangna CHINA New’~ez~EPA~ BH~rAN -977 Ohaka Hong Kong. ,T~),p~,." T~IWAN V~M (2} ~HIUPPINES SRI LA.N KA 60 KuaJa SINGAPORE PALAU ~-/ Put your company’s name on this map. For map licenses and custom mapping serwces, contact: ~- .- TeleGeography, Inc. Suite 1000, 1150 Connecticut Avenue, NW; Washington DC20036 USA TeL +I 202-467-0017; fax +I 202-467-5915 e-mail: info@telegeography.com "-~_ Unauthorized reproduction is prohibited. All rights reserved. NORTHERN MARIANAS GUAM MARSHALL ISLANDS MICRONESIA INMARSAT MOBILE SERV PACIFIC OCEAN SOLOMON ISLANDS AUSTRALIA VANUATU NEW CALEDONIA Sydney 121 Canberra, NEWZEALAND TeleGeography1996/97 © TeleGeographg, Inc. 1996 International Loa Angele: Lima Philadelphia i de Janeiro S~.o Paulo 10h A-IO © TeleGeography, Inc. 1996 Blue Pages Time Zones Lisbon aghdad 17:30 Dar es Salaam Countries and areas which have adopted the Universal Time System (UTC). Countries and areas with local deviations from the Universal Time System. 3saka A-11 TeleGeography!996/97 © TeleGeographM, Inc. 1996 X PA CIFIC /_.__ o~CEAN OCEAN A-12 © TeleGeography, Inc. 1996 Blue Pages Political Geography ARCTIC LEONE UBERIA GHANA ~AN PA C tFI C ~O,,CEAN INDIAN 0 EA N "; ................................ NEW ZE~L~ ATLANTIC OCEAN EUROPE 1. NETHERLANDS 2. BELGIUM 3. LUXEMBOURG 4. CZECH REPUBLIC 5. SLOVAK REPUBLIC 6. SWITZERLAND 7. LIECHTENSTEIN 8. AUSTRIA 9. HUNGARY 10. SLOVENIA 11. CROATIA 12. BOSNIA 13. YUGOSLAVIA 14. ALBANIA 15. MACEDONIA (F.Y.R.) 16. MOLDOVA ASIA 17. GEORGIA 18. ARMENIA 19. AZERBAIJAN 20. TURKMENISTAN 21. UZBEKISTAN 22. TAJIKISTAN 23. KYRGYZSTAN AFRICA 24. BURKINA FAS0 25. TOGO 26. EQUATORIAL GUINEA A-13 TeleGeol~raphy1996/97 © TeleGeography, Inc. 1996 National Telecommunications Indicators (A-L) Population Area 1995 (millions) (Miles2 thous.) Algeria 27.9 920.0 Argentina (a) 34.6 1068.0 Australia 18.0 2968.0 Austria 8.0 32.4 Bahrain (a) 0.6 1 Bangladesh 119.8 56.0 Belgium (a) 10.1 11.8 Bolivia (a) 7.4 424.0 Brazil 161.6 3286.0 Bulgaria 8.4 43.0 Canada (a,c) 29.5 3852.0 Chile 14.3 292.0 China 1201.0 3705.0 Colombia (a) 36.9 440.0 Croatia (a) 4.8 21.8 Cyprus 0.7 3.6 Main Lines Main Lines Lines/t00 Cellular phones Fax machines PCs 1995 (millions) 1990 (millions) people 1995 1995 (thous.) 1995(thous.) 1995 (thous.) 1.2 0.8 4.0 4.7 5.2 n.a. 5.5 3.1 16.0 340.7 n.a. 850 9.2 7.8 51.0 2304.9 450.0* 5200 3.7 3.2 47.0 383.5 n.a. 1000 0.1 0.1 24.0 27.6 5.1 n.a. n.a. 0.2 n.a. n.a. 2.0* n.a. 4.6 3.9 46.0 235.0 n.a. 1400 n.a. n.a. n.a. 7229.0 n.a. n.a. 12.0 9.4 7.0 1285.5 n.a. 2100 n.a. 2.2 n.a. n.a. n.a. n.a. n.a. 15.3 n.a. 2560.0 n.a, 5100" 1.9 0.9 13.0 197.3 n.a. 540 40.7 6.9 3.0 3629.0 270.0 2600 3.9 2.4 10.0 274.6 79.7* 630 1.3 0.8 27.0 33.7 28.4* n.a. 0.3 0.2 47.0 44.4 n.a. n.a. Czech Republic (a) 10.3 30.0 Denmark 5.2 16.6 Dominican Rep. (a) 7.8 18.8 Ecuador 11.5 109.0 Egypt 58.7 387.0 Estonia (a) 1.5 17.0 Finland 5.1 131.0 France 58.1 213.0 Germany 81.9 138.0 Greece 10.5 51.0 Hong Kong (a,b) 6.2 1 Hungary (a) 10.2 35.9 Iceland (a) 0.3 40.0 India (a,b,d) 929.3 1269.0 Indonesia (a) 193.3 735.0 Iran (a) 64.1 636.0 Ireland (a,b) 3.6 27.1 Israel (a) 5.5 10.3 Italy 56.9 116.0 Japan (b) 125.1 146.0 Jordan 4.3 34.0 Korea, Republic of (a) 44.8 38.2 Kuwait 1.6 6.9 Luxembourg 0.4 1.0 2.4 1.6 23.0 50.0 71.3 450* 3.2 2.9 61.0 819.3 n.a. 1414 n.a. 0.3 n.a. n.a. n.a. n.a. 0.7 0.5 6.0 53.0 n.a. n.a. 2.7 1.7 5.0 7.4 21.6* 194" 0.4 0.3 28.0 30.4 13.0 n.a. 2.8 2.7 55.0 1017.6 132.0 810" 32.4 28.1 56.0 1379.0 1200.0" 9300 40.4 31.9 49.0 3500.0 1446.6" 13,500 5.2 3.9 49.0 273.0 n.a. 350 3.3 2.5 53.0 763.2 284.9* 660* 1.9 1.0 19.0 265.0 n.a. 400 0.1 0.1 56.0 30.0 n.a. n.a. 11.9 5.1 1.3 135.5 50.0* 1000 3.3 1.1 2.0 218.6 55.0" 530* n.a. 2.2 n.a. n.a. 30.0* n.a. 1.3 1.0 37.0 158.0 n.a. 490* 2.3 1.6 42.0 n.a. n.a. n.a. 24.8 22.4 44.0 3863.0 n.a. 4800 61.0 54.5 49.0 10204.0 6000.0* 19,000 0.3 0.2 7.0 11.5 31.0" n.a. 18.6 13.3 41.0 1641.0 375.0" 5000* 0.4 0.3 23.0 117.6 35.0 n.a. 0.2 0.2 56.0 26.8 n.a. n.a. Source: International Telecommunication Union ©TeleGeography, Inc. 1996 A-14 © TeleCeographg, Inc. 1996 Blue Pages International Telephone Traffic Outgoing mMiTr 1994 1995 Change 94-95 1994 80.0 79.0 -1.3% n.a. 175.0 179.4 2.6% 252.6 852.0 1024.0 20.2% n.a. 819.2 901.1 10.0% 774.5 86.8 88.7 2.2% n.a. 22.1 33.0 49.4% n.a. 1049.0 1105.7 5.4% 1093.9 18.0 20.8 15.6% n.a. 199.0 319.4 60.5% 408.0 82.7 85.0 2.8% n.a. 861.2 897.9 4.3% 543.8 73.5 136.9 86.3% n.a. 1170 1390 7.3% n.a. 120.3 127.3 5.8% 302.8 185.5 210.7 13.6% 240.2 106.6 117.4 10.2% 79.0 157.6 186.8 18.5% 210.0 488.4 532.6 9.0% 500.9 63.5 85.4 34.5% 404.0 36.4 37.2 2.3% 128.6 80.0 99.8 24.8% n.a. 48.1 53.0 10.2% 50.8 259.0 315.4 21.8% 285.0 2602.5 2804.6 7.8% 2739.5 5147.1 5244.0 1.9% 3881.2 422.7 467.9 10.7% 441.2 1578.4 1691.8 7.2% 1446.4 236.6 247.5 4.6% 211.9 26.0 28.9 11.5% 25.5 314.0 341.4 8.7 % 615.0 182.5 205.9 12.8% 244.7 208.4 210.4 0.9% n.a. 323.7 407 25.7% 442.9 213.0 252.3 18.5% n.a. 1708.0 1908.2 11.7% 1864.0 1524.8 1631 6.2% 1140.6 57.0 71.7 25.7% 114.0 440.4 557 26.5% 555.2 120.6 n.a. n.a. 127.0 213.5 232.2 8.8% 145.2 Incoming mMiT~ Surplus/(Deficit) 1995 Chanqe 94-95 1994 1995 n.a. n.a. n.a. n.a. 299.4 18.5% 77.7 119.9 n.a. n,a. n.a. n.a. n.a. n.a. (44.7) n.a. 299.4 n.a. n.a. 210.6 122.1 n.a. n.a. 89.1 1172.0 7.1% 44.9 66.3 49.2 n.a. n.a. 28.4 495.5 21.4% 209.0 176.1 n.a. n.a. n.a. n.a. 603.4 11.0% (317.4) (294.5) n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. 351.5 16.1% 182.5 224.2 309.0 28.6% 54.8 98 87.3 10.4% (27.6) (30.2) 223.7 6.5% 52.4 36.9 551.0 10.0% 12.4 18.4 424.1 5.0% 340.5 338.7 154.8 20.4% 92.2 117.6 376.1 n.a. n.a. 276.3 56.0 10.2% 2.7 3.0 345.0 21.1% 26.0 29.6 2958.9 8.0% 137.0 154.3 n.a. n.a. (1,265.9) n.a. 505.4 14.5% 18.6 37.4 1598.3 10.5% (132.1) (93.5) 243.7 15.0% (24.7) (3.8) 28.4 11.2% (0.4) (0.6) 805.4 31.0% 301.0 464.0 286.0 16.9% 62.2 80.0 199.0 n.a. n.a, 11.1 n.a. n.a. 119.2 n.a, 345.6 n.a. n.a. 93.3 1999.8 7.3% 156.0 91.6 1321 15.8% (384,2) (310) 118.0 3.5% 57.0 46.3 672.0 21.0% 114.8 115 130.2 2.6% 6.4 n.a. 174.5 20.2% (68.3) (57.7) See individual country tables for carriers and routes included in outgoing and incoming traffic totals. a. International MiT[ based on billing point of traffic. b. International traffic for year ending 31 March. c. Traffic data exclude U.S. route. Traffic data exclude some carriers or routes. (See country table for details.) Data for 1994. Source: TeleGeography, Inc. ©TeleGeography, Inc. 1996 A-15 TeleGeo{~raphy 1996/97 © TeleGeography, Inc. 1996 National Telecommunications Indicators (M-Z) Population Area Main Lines Main Lines Lines/100 Cellular phones Fax machines PCs 1995 (millions) (Miles2 thous.) 1995 (millions) 1990 (millions) people 1995 1995 (thous,) 1995 (thous.) 1995 (thous.) Macau (a) 0.4 1 0.1 0.1 37.0 30.6 7.2* n.a. Macedonia, TFYR (a) n.a. 10.0 n.a. 0.3 n.a. n.a. 1.8" n.a. Malaysia (b,d) 20.1 127.0 3.3 1.6 17.0 872.8 571.7" 640* Mexico (a) 91.8 756.0 8.8 5.4 10.0 642.0 n.a. 2400 Moldova 4.3 13.0 0.6 0.5 13.0 0.0 n.a. n.a. Morocco (a) 26.9 172.0 1.1 0.4 4.0 29.5 13.8" n.a. Netherlands (a) 15.5 16.2 8.1 6.9 53.0 513.0 n.a. 3100 New Zealand (b) 3.5 104.0 1.7 1.5 49.0 388.0 50.0* 669* Norway (a) 4.3 125.0 2.4 2.1 56.0 982.0 n.a. 1193 Pakistan 129.7 307.0 n.a. 0.8 n.a. n,a. 8.0* n.a. Paraguay 4.9 157.0 n.a. 0.1 n.a. 15.8 n.a. n.a. Peru (a) 23.7 496.0 1.1 0.6 5.0 73.5 n.a. n.a. Philippines 67.5 116.0 1.4 0.6 2.0 492.7 35.0" n.a. Poland (a) 38.6 121.0 5.7 3.3 15.0 75.0 n.a. 1100 Portugal 9.9 35.7 3.6 2.4 36.0 340.0 n.a. 600 Qatar 0.5 4.2 0.1 0.1 27.0 18.5 9.8* n.a. Russia (d) 147.3 6592.0 25.0 20.7 17.0 88.5 n.a. 2600 Saudi Arabia 17.9 830.0 1.9 1.2 11.0 16.0 n.a. n.a. Singapore (b) 2.9 1 1.4 1,1 47.0 291.8 n.a. 430* Slovak Republic 5.4 19.0 1.1 0.7 21.0 12.3 37.9* n.a. Slovenia 2.9 7.8 0.6 0.4 31.0 27.0 15.5 n.a. South Africa 41.4 471.0 3.9 3.3 9.0 535.0 340.0* 875* Spain 39.2 195.0 15.1 12.6 39.0 944.0 n.a. 3200 Sri Lanka (a) 18.1 25.3 0.2 0.1 1.0 51.3 30.0* n.a. Sweden (a) 8.8 174.0 6.0 5.8 68.0 2025.0 n.a. 1700 Switzerland (a) 7.0 15.9 4.3 3.9 61.0 447.2 197.0 2450 Syria 14.6 71.0 0.9 0.5 6.0 n.a. 5.0 n.a. Taiwan (a,b) 21.3 13.9 9.2 6.3 43.0 770.4 430.0* 1773 Thailand 58.7 198.0 3.5 1.3 6.0 1087.5 60.0" 680* Tunisia 8.9 63.0 0,5 0.3 6.0 3.2 2.7 44* Turkey 61.9 301.0 13.2 6.9 21.0 436.5 87.6 n.a. United Arab Emirates 2.4 32.3 0.7 0.4 29.0 128.5 30.4 n.a. United Kingdom (b,d) 58.5 94.0 n.a. 25.8 n.a. 5737.0 n.a. 10,900 United States (a) 263.1 3619.0 165.0 136.3 63.0 33785.6 14052.0" 86,300 Uruguay (a) 3.2 68.0 0.6 0.4 20.0 40.0 n.a. n.a. Venezuela (a) 21.7 352.0 2.5 1.5 11.0 n.a. n.a. 370 Vietnam 73.5 127.0 0.4* n.a. 17.4" 13.2" 7* n.a. Yugoslavia 10.5 39.5 2.0 1.7 19.0 n.a. 15.0 n.a. Source: International Telecommunication Union ©TeleGeography, Inc. 1996 A-16 © TeleGeography, Inc. 1996 Blue Pal~es International Telephone Traffic Outgoing mMiTr 1994 1995 Change 94-95 1994 100.0 108.1 8.1% 84.3 35.1 35.5 1.3% 78.3 342.3 408.3 19.3% 399.7 844,1 950.0 12.5% 1829.4 73.9 66.0 -10.6% n.a. 130 129.7 -0.1% n.a. 1345.8 1458.7 8.4% 1290.9 261 312.0 35.7% 263 395.5 431.5 9.1% 352,0 61.4 65.9 7.4% n.a. 18.1 19.4 7.2% 30.6 51.0 66.7 30.8% 178.6 160.0 174.0 8.7% 617.0 356.6 381.4 7.0% 643.8 262.4 283.9 8.2% 467.8 62.7 75.8 20.9% n.a. 229.2 287.4 25.4% 365.0 477 499.1 0.0% n.a. 643.0 773.0 20.2% n.a. 52.5 58.8 12.1% 68.5 90.6 100.6 11.0% 83.2 262.6 280.0 6.6% n.a. 948.3 1024.6 8.0% 969.9 23.7 27.5 16.0% 78.7 802.0 900.0 12.2% n.a. 1649.3 1778.4 7.8% 1353.0 40.0 66 65% 78.0 498.5 592.8 18.9% 613.5 173.2 218.8 26.3% 313.3 64.0 69.9 9.2% n.a. 284.3 373.6 31.4% 601.4 428.2 503.6 17.6% n.a. 3507.0 4016.0 14.5% 3577.0 13200.3 15623.0 18.4% 6133.1 46.3 49.9 7.8% 67.7 141.3 129.1 -8.6% 164.3 24.0 35.1 46.4% n.a. 181.9 212.8 17.0% 229.0 Incoming mMiTr Surplus/(Deficit) 1995 Chan=je 94-95 1994 1995 90.4 7.2% (15.7) (17.7) 74.0 -5.5% 43.2 38.5 442.0 10.6% 57.4 33.7 2114.0 15.6% 985.3 1,164.0 n.a. n.a. n.a. n.a. 265.0 n.a. n.a. 135.3 1453.0 12.6% (54.9) (5.7) 327.0 35.7% 2.0 15.0 373.2 6.0% (43.5) (58.3) 362.1 n.a. n.a. 296.1 n.a. n.a. 12.5 n.a. 195.4 9.4% 127.6 128.7 691.0 12.0% 457.0 517.0 649.3 0.9% 287.2 267.9 525.0 12.2% 205.4 241.1 52.6 n.a. n.a. (23.2) 448.1 22.8% 135.8 160.7 n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. 81.6 19.2% 16.0 22.8 121.2 45.6% (7.4) 20.6 B.a. n.a. n.a. n.a. 1076.4 11.0% 21.6 51.8 92.0 18.9% 55.0 64.5 n.a. n.a. n.a. n.a. 1439.3 6.4% (296.3) (339.1) n.a. n.a. 38.0 n.a. 545.3 -11.1% 115.0 (47.5) 277.7 -11.4% 140.1 58.9 n.a. n.a. n.a. n.a. 705.0 17.2% 317.1 331.5 n.a. n.a, n.a. n.a. 4021.0 12.4% 70.0 5.0 7010.6 14.3% (9,502.3) (8,623.0) 73.9 9.2% 21.4 24.0 186.6 13.6% 23.0 57.4 170.3 n.a. n.a. 135.2 296.0 29.3% 47.1 83.2 See individual country tables for carriers and routes included in outgoing and incoming traffic totals. a. International MiFr based on billing point of traffic. b. International traffic for year ending 31 March. c. Traffic data exclude U.S. route. d. Traffic data exclude some carriers or routes. (See country table for details.) Data for 1994. Source: TeleGeography, Inc. ©TeleGeography, Inc. 1996 A-17 For detailed information on other TeleGeography, Inc. publications, contact: TeleGeography, Inc. Suite 1000 1150 Connecticut Avenue, NW Washington, DC 20036 USA Tel.: ÷ 1 (202) 467-0017 Fax: ÷ 1 (202) 467-5915 E-mail: sales@telegeography.com Or visit our web site at: http ://www.te I e g eog ra phy. c om Telecommunications Map of the World TeleCeography, Inc./Petroleum Economist, May 1996 Submarine cable capacity, satellite locations, traffic flows and over 400 national and city dialing codes make this full color poster a practical business tool as well as an attractive wall decoration. Border illustrations detail the world’s top carriers, national traffic balances, cellular vs. wireline subscribers by country, Internet backbone networks, satellite footprints and the installed base of various communication terminals. Size is approx. 32" x 54" (.9 m x 1.4 m). Map is shipped folded. ISBN 1-86186-005-6 Direction of Traffic 1996 TeleCeography, lnc./ITU, November1996 With nine years (1986-1994) of traffic data and five years of tariff data for 55 countries, Direction of Traffic I996 is the definitive study of recent trends in international telecommuni- cations. In addition to the statistics, a comprehensive analysis (paper edition only) answers some of the most basic and yet thorniest questions about the pricing and volume of interna- tional calls: ¯ In an era of rapid technological advance, why do calls still cost so much? ¯ How low can international tariffs go? ¯ How much does it really cost to complete an international call? ¯ What are the effects of resale, call-back, Intemet telephony, and other new regimes? As the old regime of monopoly carriers and half-circuits collapses, this report sketches the new world. 200 pages. ISBN 92-61-06291-1. Direction of Traffic 1996 Supplemental Diskette A supplemental diskette contains even more detailed route-by-mute traffic breakdowns for over 100 countries. Pick a set of orig- inating and terminating countries, choose a time series, and then with one click the data is ready for analysis. Add one more step and the data can be exported to Excel, Quattro or Lotus. Requirements: MS DOS 2.1 (or Windows). This version does not have the essay, forecasts, or traffic totals found in the printed version. The TeleGeography 1 O0 TeleGeography, Inc., May 1996 "It is an essential reference for anybody hoping to understand the bewil- dering tangle of alliances and partnerships which characterise the elec- tronics industries today. "--Alan Cane, The Financial Times The TeleGeography 100 An up-to-date guide on who owns what in the global information economy. Specially researched ownership charts for over 100 top entertainment, telecoms, computer and equipment companies make sense of the hundreds of mergers, demergers, acquisitions, joint ventures and alliances which characterize today’s,~’~ ~-~ digital marketplace. The report covers all the main telco and cable TV mergers trig- gered by the new U.S. communications law. See at a glance why the world’s largest auto manufacturer is also a leading player in satellites and computer systems. Track the expanding reach of telco alliances, such as WorldPartners, Concert and Global One. See how Internet projects link tel- cos, software companies, broadcasters and tel- cos worldwide. Charts show all major corporate subsidiaries and affiliates as well as allies and partners. There is also a comprehensive cross index covering more than 2000 companies iden- tiffed in the report. Contact information for the TeleGeography 100, annual revenues, key exec- utives and a precis of each company’s strategic significance are also provided. 247 pages. ISBN 1-886142-06-8. New International Carriers 1996 TeleGeography, Inc., September 1996 The first (and only) comprehensive directory of the world’s newest international telephone companies has contact information on over 500 carriers and prospec- tive market entrants in 33 countries, including the U.S., Canada, Mexico, Chile, the U.K., Germany, Russia, Japan, Australia, and the Philippines. The directory cov- ers all new international facilities-based telephone card- ers, announced market entrants, globa! mobile satellite companies, ISR carriers, private satellite and cable com- panies, and global telephone alliances, as well as IVAN providers in Japan and Hong Kong. Addresses, tele- phone/fax numbers and contact names are included for each carrier. Major stockho!ders and affiliates are also listed (with contact details). A special charts section profiles the ownership interests of important new carriers as well as the global telephone alliances. Market entry rules and national regulators are also listed. 200 pages. ISBN 1-886142-05-X. Competitive markets included: Australia Denmark Hong Kong Korea Portugal Belgium Dominican Republic Indonesia Malaysia Russia Bermuda Finland Ireland Mexico South Africa Canada France Israel Netherlands Spain Chile Germany Italy New Zealand Sweden China Ghana Japan Philippines Switzerland Uganda United K~ngdom United States TeleGeography t 996/97 "’City of darkness.’ Between the walls of the world."... "It’s a MUD, right?" Something like a larger, permanent version of the site the Tokyo chapter had erected for the meeting, or the tropical forest Kelsey and Zona had put up. But people played games in MUDs; they made up characters for themselves and pretended. Little kids did it, and lonely people. "No," he said, "not a game." They were inside now, smoothly accelerating, and the squirming denM sity of the thing was continual visual impact, an optical drumming. "Tai Chang Street." Walls scrawled and crawling with scrolling messages, spectral doorways passing like cards in a shuffled deck .... Fractal filth, bit-rot, the corridor of their passage tented with crazy swoops of faintly flickering lines of some kind. ’~ms House Backstreet." A sharp turn. Another. Then they were ascending a maze of twisting stairwells, still accelerating, and Chia took a deep breath and closed her eyes .... when she opened [them] they were in a much cleaner but no larger version of [Masahiko’s] room behind the kitchen in the restaurant... Chia looked around at the reproduction of his tiny room. "Why don’t you have a bigger site?" Instantly worried that it was because he was Japanese, and maybe they were just used to that. But still it was about the smallest virtual space she could remember having been in, and it wasn’t like a bigger one cost more, not unless you were like Zona and wanted yourself a whole country. "The Walled City is a concept of scale. Very important. Scale is place, yes? Thirty-three thousand people inhabited original. Two-point-seven hectares. As many as fourteen stories." None of which made any sense to Chia. "I have to port, okay." "Of course," he said, and gestured toward her Sandbenders..." The bit-mapped fish swam back and forth .... She looked at the door to her bedroom and found herself wondering what she’d find there if she gestured for it .... She looked at her stack of Lo/Rez albums beside the lithographed lunch box, her virtual Venice beside that. Even her Music Master would seem like company now. She opened it, watching the Piazza decompress like some incredibly intricate paper pop-up book on fast-forward, facades and colonnades springing up around her, with the hour before a winter’s dawn for backlight. Turning from the water, where the prows of black gondolas bobbed like marks in some lost system of musical notation, she lifted her finger and shot forward into the maze, thinking as she did that this place had been as strange, in its way, as Masahiko’s Walled City, and what was that all supposed to be about anyway?.... Chia closed her eyes and counted to three. Made herself feel the carpeted floor she sat on in the Hotel Di. She opened her eyes. At the end of the narrow Venetian street, down the tilted, stepped cobbles, where it opened out into a small square or plaza, an unfamiliar figure stood beside the central fountain. She pulled the goggles off without bothering to close Venice. Masahiko sat opposite her, his legs crossed, the black cups sucked up against his eyes. His lips were moving, silently, and his hands, on his knees, in their black tip-sets, traced tiny finger-patters in the air...." Excerpted from Idoru (G. P. Putnam’s Sons, New York, 1996) Copyright © 1996 by William Gibson TeleGeography 996/97 Gibson’s Geographica setting is on-line or off, the striking landscapes in William Gibson’s books pro- vide a compelling geographical vision of tomorrow. Since the first novel, Neuromancer (1984), which introduced the world to cyberspace--a word Gibson coined--his work has been marked by a unique attention to place. Though often (mis)labeled a science fiction writer, Gibson’s ability to invest everyday settings with a life force of their own has more in common with such early modern stylists as Joseph Conrad and Franz Kafka. The places which Gibson animates are decidedly post-modern, however. Gibson has also mastered the art of simultaneously fast-forwarding many different domains--the physical and the virtual, the econom- ic and the social. For example, in Virtual Light (1993) and Idoru (1996) we learn that most large countries have long ago splintered into private economic zones and metroplexes. Earthquakes have shattered San Francisco and Tokyo. The seas have flooded Venice. And personal "home pages" have been transformed into 3-D homes. The following excerpt from ldoru, introduces some of Gibson’s geographica, including a new city of bits-- Hak Nam, the Walled City--"of the Net, but not on it." Chia, a fourteen year old from Seatac (Seattle/Tacoma) has been sent to Japan by the Lo/Rez fan club to find our whether the lead singer, Rez, is really in love with Rei Toei, a beguiling software projection who has captivated local audiences. Her guide to the intrigue surrounding Rez’s near fatal attraction is sev- enteen year old Masahiko. Chia is staying with his sister. Chia and Masahiko sat facing one another on the white carpet .... Chia had her Sandbenders across her knees and was working her fingers into her tip-sets. Masahiko’s computer was on the carpet in front of him; he’d put its control-face back on and peeled a very compact pair of tip-sets out of the back of the cube, along with two small black ova] cups on fine lengths of optical cable. Another length of the cable ran from his computer to a smalt open hatch at the back of the Sandbenders. "Okay," Chia said, settling the last of her tips, "let’s go. I’ve got to get hold of somebody..." "Yes," he said. He picked up the black cups, one in either hand, and placed them over his eyes... Chia reached up and pulled her own glasses down, over her eyes. "What do I--" Something at the core of things moved simultaneously in mutually impossible directions, it wasn’t even like porting. Software conflict? Faint impression of light through a fluttering of rags. And then the thing before her: building or biomass or cliff face looming there, in countless unplanned strata, nothing about it even or regular. Accreted patchwork of shallow random balconies, thousands of small windows throwing back blank silver rectangles of fog. Stretching either way to the periphery of vision, and on the high, uneven crest of that ragged facade, a black fur of twisted pipe, antennas sagging under vine growth of cable. And past this scribbled border, a sky where colors crawled like gasoline on water. "Hak Nam," he said, beside her. "What is it?" (over)