Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
APPENDIX A 237 APPENDIX A Federal Networking: The Path to the Internet DEVELOPMENT AND GROWTH OF THE INTERNETâ A THUMBNAIL SKETCH The Federal Networking Legacy Federal networking activities, in the context of this report, originated with the ARPANET program, itself the seed for the larger and more amorphous Internet, which in turn has provided a foundation for the National Research and Education Network (NREN) program. Beginning in the late 1960s the U.S. Defense Department's Advanced Research Projects Agency (ARPA) funded the high-risk, high-payoff development of the ARPANET, a 56-kbps backbone network.1 The first ARPANET node was installed at the University of California at Los Angeles in September 1969, thus launching the first packet switching network; by 1971 approximately 20 nodes had been installed, and ARPA was funding 30 different university sites as part of the ARPANET program. In the mid-1970s, the TCP/IP protocols were developed to link together different packet networks. In the late 1970s and early 1980s, research versions of local area networks and workstations were connected to the ARPANET, thus forming what is now more widely known as the Internet. At the time of the transition from the original ARPANET protocol, the Network Communication Protocol, to TCP in 1983, only a few hundred computers were on the nascent Internet, which connected only a handful of networks.2 Use of the ARPANET was expanded in the 1970s to the computer science research community and to segments of the science research community supported by the federal mission-oriented agencies. As a result of this expanded use and more regular operations, in July 1975 ARPA
APPENDIX A 238 transferred ARPANET management (previously the responsibility of Bolt, Beranek, and Newman) as well as the Network Measurement Center (previously the responsibility of UCLA) to the Defense Communications Agency (DCA; now the Defense Information Systems Agency) with the expectation that direct experience with packet switching by DCA would ultimately be of wider benefit to the Department of Defense. Meanwhile, during the late 1970s and early 1980s, a number of federally supported, discipline- specific networks were established. Among these was MFEnet, which was funded by the Department of Energy (DOE) to give academic physicists doing research in nuclear fusion access to supercomputers at the Lawrence Livermore National Laboratory. MFEnet specified and implemented its own communications protocols. Two other federally supported networks were the DOE-funded HEPnet (for high-energy physics research) and the Space Physics Analysis Network (SPAN) funded by the National Aeronautics and Space Administration (NASA). SPAN and HEPnet were part of a global DECnet- based network that supported collaboration by scientists working in the space sciences and physics. The National Science Foundation (NSF) has played a critical role in the second decade of research networking's evolution, funding research into relevant technologies as well as network deployment and use. In the period from 1980 to 1986, NSF supported the development of CSNET, a "logical network" for computer science researchers. CSNET was a network of networks, one component of which used the Internet protocols over an X.25 public data network. It also included the ARPANET and PHONENET, a telephone-based electronic mail relaying system. By 1985, CSNET had links to over 170 university, industrial, and government research organizations and numerous gateways to networks in other countries. In 1987 CSNET merged with BITNET, a network serving users from academic institutions that was initiated in 1980-1981. CSNET operations were continued under the Corporation for Research and Education Networking, whose operating costs were completely covered by member organizations' dues. Its mission apparently accomplished, CSNET service was discontinued in the fall of 1991.3 In 1985, NSF initiated a program of networking and computer support for centers with supercomputers to be used by researchers across the science and engineering research community. This program began with a memorandum of understanding with ARPA to allow NSF-funded supercomputer centers and selected researchers to use the ARPANET. NSF instituted the NSF Connections program in 1986 to broaden the base of network users with their own computer facilities and eventually to help universities achieve access to supercomputers (by supplying supporting hardware and/or telecommunications lines for direct, point-to-point connections); also in 1986, it launched the NSFNET network backbone pro
APPENDIX A 239 gram. After significant congestion was experienced in 1987, the backbone was upgraded to T1 service (1.5 Mbps) that became operational in 1988. Between 1986 and 1987, the point-to-point Connections program grew to include support for more complete networking arrangements that would collect network traffic and provide access to the limited NSFNET. That period began with funding for SURAnet and NYSERNet regional network proposals and saw over a dozen regional networks come into being (e.g., BARRnet, Midnet, PSCnet, Sesquinet, and Westnet).4 The Connections program evolved to emphasize connecting universities and other research and educational institutions to the Internet, and under the new very high speed backbone network (vBNS) orientation (see "Continuing Evolution of Provisions," below) it is expected to support institutional needs for vBNS access (with awards based on evaluation of competing proposals). The NSF's funding arrangements have given rise to a three-tier structure of campus (primarily universities and research institutions), regional, and backbone networks serving the research community (see Figure 1.2 in Chapter 1).5 This conceptualization is complemented by commercial networks, international networks, and other interconnections that have resulted in a globally interconnected mesh of networks known as the Internet.6 NSFNETâThe Research and Education Communities' Link to the Internet The late 1980s witnessed what might be called a rationalization of federal networking activities of the previous two decades. Initial hardware, software, and reliability problems among regional networks helped to motivate the development of network operations centers; such a center became an element of the 1987 NSFNET backbone solicitation. According to one characterization, "NSFNET has a very special role in the [Internet] hierarchy: it acts as a generic transit, routing, and switching network for research and education networks."7 Synergistic Growth of NSFNET and the Internet The concept of what the NSFNET could be and recognition of the problems it posed for parties operating and using it expanded steadily through the late 1980s and early 1990s. The backbone network speed was upgraded from 56 kbps to T1 and then T3; this backbone (45-Mbps) network grew to a size of 19 nodes, including 16 sponsored by NSF and 3 serving interagency communication needs. In 1989, 20 years after its birth at UCLA, the ARPANET was officially decommissioned; its descen
APPENDIX A 240 dent, the NSFNET, inherited its role as the research and education communities' backbone network. Through the NSFNET program NSF spurred the growth of the Internet and provided a national demonstration of the feasibility of the technology and the market potential for open data networking. NSF also nurtured the broadening of the community served by the NSFNET through its requirements that the supercomputer centers and regional networks it supported develop their markets and move toward self-sufficiency; the phased withdrawal of financial support for regional networks, now linked to the termination of support for the NSFNET backbone, is the latest manifestation of NSF's promotion of self- sufficiency in the dispersed network-based activity it has fostered. Continuing Evolution of Provisions for Research and Education Networking As discussed in Chapter 5, the shift from direct provision of broad-based backbone network service through the NSFNET backbone to an expectation that commodity backbone and interregional networking services will be procured directly by users or their institutions is another part of a pattern of emphasis on eventual self-sufficiency. The new NSFNET, circa 1994, will involve a four-node (at least initially), very high speed backbone network service (vBNS), with network access points (NAPs) for connections among various networks serving the research and education communities and others (see Box 1.3 in Chapter 1). The NAPs themselves may eventually be turned over to participating networks for funding as well as management, much as commodity backbone transport is being handed off imminently.8 NSF has indicated that support for connection to the vBNS is expected to be linked to proposals for expanding enhanced capabilities provided by regional networks, which will effectively serve as capillaries for an otherwise small vBNS.9 The National Research and Education Network Programâ Expansion from the Internet Base and Earlier ARPANET Efforts The NREN program emerged during the period from 1987 to 1989 as part of the launch of the High Performance Computing program (now known as the High Performance Computing and Communications (HPCC) initiative). Within the NREN framework, different federal agencies (notably, NSF, NASA, DOE, and ARPA) launched or expanded separate but interconnected networking efforts; these became components of the Internet.10 These networks resulted from the early understanding by the four dominant NREN agencies of the potential value of such services to the researchers that they fund.11
APPENDIX A 241 Each agency backbone network utilizes packet switching technology developed by the ARPANET as well as the results of early Internet projects at universities and national laboratories plus complementary research in industry; many of these projects fueled commercial spin-offs and business development. The ARPA packet radio program, for example, supported development of the protocols that are the basis of the Internet; early protocol developers used the ARPANET to support their collaboration. The earliest TCP protocol work was undertaken in the mid-1970s at Stanford University, Bolt, Beranek, and Newman (BBN), and University College in London; the Internet work was thus international from the outset. The ARPANET also inspired a number of commercial data networks (such as Telenet) in the mid-1970s. Mission agency networks serve specific communities: space and earth scientists in the case of the NASA Science Internet; scientists doing energy- related research in the case of ESnet; defense-related researchers and others in the case of the various networks of ARPA, including DARTnet and especially the Terrestrial Wideband Network TWBnet; and the broadest assortment of scientists and disciplines in the case of NSFNET. Some agencies, such as the National Institutes of Health and within it the National Library of Medicine (NLM), have launched network-based information services that depend on additional commercial networks for access.12 The four principal HPCC agencies also support the development of new and enhanced networking and related technologies under the NREN umbrella. ARPA, for example, has emphasized scalable architectures as well as high- speed computing and networking, exploring concepts and implementation issues with NSF through the gigabit testbed activities and now also through the optical network program.13 DOE and NASA support development of "enabling technologies" as well as applications in support of their mission needs. For example, DOE has contributed to the CASA gigabit testbed through Los Alamos National Laboratory efforts. This work supported DOE's mission interest in high-speed networking and progress in global climate modeling, quantum chemical reaction dynamics, and three-dimensional seismic profiling. Growth of Internet Use and Development Beyond the Research Community Connectivity for the research community has driven the development of the Internet, but over the past five years access by the K-12 education and library communities has grown more rapidly, aided in part by outreach programs from the research community and the HPCC funding agencies. DOE, NASA, and the National Oceanic and Atmospheric Ad
APPENDIX A 242 ministration (NOAA), for example, have begun to package and provide access to scientific data and tools in programs intended for K-12 educational applications. NSF has also supported the broadening of Internet access through pilot projects for K-12 networking support through the Education and Human Resources Directorate and through pilot projects for library networking supported by the Department of Education's Title II-D program. Finally, HPCC agencies have also reached out to other, specialized communities through specific applications and services. The NLM's programs aimed at health care researchers and parties within the health care delivery system are an obvious and important example. Internet product development has also grown significantly since 1980, building on federally funded research. In addition to the commercial spread of the TCP/IP protocols, a related development was the incorporation of the UNIX operating system into local area network (LAN) and high-performance workstation products, building on efforts supported by ARPA at BBN and at the University of California at Berkeley. In the 1990s, growth in commercial activity has been most evident for communications and information services, growth for which NSFNET has been a catalyst. For example, PSI was spun off from NSYERNet; ANS was an outgrowth of the Merit regional networking activities; CERFnet was launched by General Atomics, which also runs the NSF-funded San Diego Supercomputer Center. 14 Around 1990 a conscious effort was made to link commercial and nonprofit information service providers such as Dow Jones, Dialog, NLM, and CARL to the Internet. Transition to a New Era in Networking The latest federal initiative relating to networking is the expansion of the HPCC to include the new Information Infrastructure Technology and Applications (IITA) program. This direction was made possible by the High- Performance Computing Act of 1991 (PL 102-194), although specific funding authorization did not come until 1994. Agency program activity under IITA commenced in 1993, building on the generation of ideas relating to so-called National Challenge areas (manufacturing, education, health, and libraries). According to an official description of IITA activities, the Gigabit Testbed program will continue to foster advanced technologies and their integration in support of prototyping for national challenge efforts. Activities will fall into four broad areas: information infrastructure services; systems development and support environments; intelligent interfaces; and national challenges (apparently, specific applications).15 At this time it is not known howâin terms of both complementarity and substitutionâthe growth of IITA will affect activities under the NREN umbrella.
APPENDIX A 243 THE INTERNET TODAYâWHAT IT IS AND HOW IT WORKS The Internet, a global interconnection of computer networks spanning 61 countries, over 2 million host computers, and an estimated 15 million users, provides paths for communication among these computers. This interconnection permits a range of activities to be accomplishedâamong them exchange of electronic mail (e-mail), exchange of files, and remote login to computersâand provides access to a growing array of on-line information. Used today by many different communities in support of collaboration, cooperation, and dissemination of information, the Internet is viewed by its creators as a public resource. The Internet is an open network. Any individual or organization is welcome to attach and become a part of the Internet. All that is required is a terminal or a computer with the correct software and the ability to pay the costs. The necessary software is now available to attach nearly all kinds of computers, including personal computers, and the options for attaching are expanding. For the most part, there are no limitations on the purposes for which the network may be used.16 Features and Use of the Internet The Internet's Services The most popular service of the Internet has been e-mail. It exists in several forms, including mail between individual users and mail to "newsgroups," which are ongoing group discussions on many different topics. Although mail and newsgroups may dominate Internet usage today, this pattern is changing with the rapid rise in applications that provide access to information (Figure A.1). Tools such as World-Wide Web (and its most popular user interface, Mosaic) are transforming the use of the Internet. Chapter 2 includes information on these and other new Internet services. Another new service now in experimental use but not yet in production is multimedia teleconferencing. In addition, audio, video, and shared work-space tools are now emerging from the research community, and content is coming from a variety of multimedia information providers. The coming availability of such services and the expected wide deployment of the next generation of very powerful workstations will have considerable impact on the required capacity and resultant costs of the Internet. (Chapter 5 discusses in some detail the issues of costs and pricing for use of the Internet.)
APPENDIX A FIGURE A.1 Internet services: global information space. Courtesy of the Internet Society, Reston, Va. 244
APPENDIX A 245 How Can One Attach to the Internet? Most users of the Internet today are affiliated with institutions. The computer they use is attached to a LAN, to which other end users are also attached; this LAN, in turn, is connected by a router to a regional or wide area network. For users located in homes or at isolated business sites, there are two ways to attach. One is to use a terminal to connect to a computer (host) that is already attached to the Internetâa very common approach to attachment today. The other is to equip the local computer with Internet software and connect not to a remote host, but to an Internet router. This mode of connection, although somewhat more complex (and more expensive), permits the local computer direct access to a fuller range of services, for example file transfer. It is also possible to use a dial-up connection to pass packets into the Internet; two of the common protocols for this purpose are SLIP and PPP, which are often mentioned in software packages offered for personal computers. The limited bandwidth of dial-up lines may constrain the use of some Internet services such as teleconferencing, but there are today few cost- effective alternatives at higher speeds. This issue is discussed further in Chapter 5. What Does It Mean to Be a Part of the Internet? To connect to the Internet, an end node must run software that implements what are called protocols. The various protocols of the Internet define the packet formats, the rules for packet exchange, the higher-level services available, such as forwarding of e-mail, and so on. The Internet protocols are often known by the names of the most important two, TCP and IP. Many computers now come with the software for Internet connection built in, and such software is commercially available for essentially any other computer today. Computers in the IBM PC class, the Macintosh, and a range of UNIX machines are all very common end nodes on the Internet. Because the Internet is a somewhat amorphous collection of networks and end nodes, the question of what it means to attach to or be a part of the Internet can cause confusion. At the technical level, an end node is a part of the Internet if (1) it implements the required protocols (the Internet protocols, including IP and TCP), (2) it has an Internet address that includes a network number known in the routing tables in the Internet routers, and (3) it has some sort of communications connection that permits it to exchange network information elements (IP packets) with the other 2 million machines on the Internet. An alternative to connecting to the Internet at the low (packet) level
APPENDIX A 246 is to connect at a higher (application) level. Even without packet-level connectivity, higher-level services can be interconnected through some sort of gateway. Thus, for example, interconnected mail systems permit the exchange of e-mail among a pool of users that is much larger than the group of users who have IP-connected hosts for mail exchange. Although users interconnected through a relatively high-level (application) gateway may feel that they are "on the Internet," their perspective is limited; as new services become available over the Internet, mail-only users will discover that such services are not accessible to them. Of the large private corporate networks, some have been implemented using the Internet protocols, and some using other, vendor-specific protocols. Because of concerns about possible breaches of security, these networks are not usually connected directly to the Internet by a packet router. Instead, a higher- level service gateway is more commonly used, which does permit mail interoperation but may limit other services. Chapter 2 discusses the importance of designing a network architecture so as to provide users a range of choices in level of connectivity to the network, and it addresses issues of network security and its relation to packet routers and higher-level service gateways. Internal Components of the Internet The typical user may be satisfied to know that the Internet is a system capable of moving data from one place to another as requested by the attached computers. But within the Internet is the wide range of technologies from which it is built. The core technical concept of the Internet is packet switching, which was proposed about 30 years ago as a very efficient way of sharing very expensive long-distance telecommunications circuits. A packet (a small amount of data with a destination address on the front) can be put in line with packets from other hosts and sent in turn down a link. This very fine grained sharing allows the cost of the link to be divided among many users. Internally, the Internet is composed of networks, almost 20,000 of them. A network is a logical entity that may in turn be composed of a number of physical network elements called subnetworks. Some of these networks are collections of LANs at campus locations, for example, Ethernets and Token Rings; some are long-distance networks, such as the NSFNET backbone network; and some are regional or state-level networks. Each network has a unique network number that is part of the address of the end nodes, or computers, attached to each net. Addresses are organized in a nested manner and are represented by four elements,
APPENDIX A 247 which are written as four numbers separated by dots. Sometimes more than one element is used to name a network or a subnetâif we used just one field to name networks, we could name only 256 nets, because each field holds only 256 values. The address of a particular machine at the Massachusetts Institute of Technology, the one on which much of this appendix was written, is 22.214.171.124. This means that the machine is host 108 on subnetwork 23.0 on network 18, with network 18 covering all of MIT and subnet 23.0 covering the building in which the end node (this author's computer) is located. The device that connects networks together and passes packets among them is called a router. Internet routers are now commercial products, and many thousands of routers currently exist within the Internet. As the name indicates, routers deal with routing, addressing, management of telecommunications links, and other issues of operation. In particular, routers maintain routing tables, which indicate for each of the known network numbers how to reach that net. At the end of March 1994 there were 19,373 networks active in the Internet.17 With the growth of the Internet, routing tables have become larger, and maintaining them in a consistent state is now becoming more challenging. Routing tables are maintained through a combination of on-line automatic message exchange and manual control. In early 1994 NSF announced an award for an Internet routing arbiter project intended to facilitate the logical interconnection of the networks attached to the Internet.18 The design of the telecommunications technology out of which the Internet is built has matured over the last decade. When the Internet was first conceived, the only wide area technology available consisted of simple point-to-point telecommunications links operating at various speeds. Now, a range of very sophisticated technologies exists for passing packets between routers, including Frame Relay, SMDS, asynchronous transfer mode, and so on. However, this increase in sophistication in the internal components of the Internet has not changed its basic nature as perceived by the end user, except that the new technologies are now facilitating provision of a broader range of services, such as real-time video, and are permitting broader and less expensive deployment of services. National and Global Interconnection of Internet Users No single organization owns or operates the Internet. Instead, several thousand organizations separately operate and administer their individual parts, which combine to form the total Internet (Figure A.2).19 Most countries have some sort of backbone or long-distance network
APPENDIX A FIGURE A.2 Growth of registered IP networks, 1989 to 1993. Graph courtesy of the Internet Society, Reston, Va. 248
APPENDIX A 249 spanning the country. The United States has a number of such "nets," some provided by the government and some built for commercial use. Attached to these backbone networks are more localized networks that serve regions, states, or cities. Finally, there are the networks that cover an institutionâa university, corporation, or similar facilityâand there are networks that permit individuals to connect from homes and businesses. Federal agencies have addressed the problem of interaction and interconnection within the United States through the concept of exchange points, first implemented as the Federal Internet Exchanges (FIXes; including FIX-West at NASA Ames Research Center and FIX-East at the University of Maryland, College Park). The multiple paths between the FIXes are individually subject to each agency's policy-based constraints on usage. The Commercial Internet Exchange (CIX) added a third, similar exchange point for commercial Internet access providers, and the network access points outlined in the new NSF network program solicitation (see Box 1.3 in Chapter 1) will further generalize the implementation of exchange points. Experience with and expectations for these network exchanges are motivating further work in the area of routing funded by NSF as part of its networking program.20 Access to and interconnection among the pieces of the Internet is offered through common carriers, leased-line access providers, and dial-up access providers, many of which are interconnected. For example, Metropolitan Fiber Systems operates fiber links between Falls Church, Virginia, and College Park, Maryland, that are known as Metropolitan Area Ethernet or MAE-East and that directly link Sprint, PSI, UUNET, NSFNET, and the main data line to Europe. Sprint leases circuits to UUNET, America Online, Digital Express, and other providers of Internet access; MCI has an equity interest in ANS and leases lines to Prodigy. Dial-up access is available from providers that range from residence- based small businesses with a single computer, modem, and telephone line to corporate communications concerns (e.g., GE Information Services or America Online).21 Provision of access to the Internet can be and has been launched as a very small business as well as a big business; one entrepreneur reported that his total hardware cost for starting a 16-telephone-line service was under $15,000.22 The Internet's very decentralized administration succeeds because it is governed by a set of defining conventions (the Internet protocols and the agreed- to operating rules) that specify what each operator must do to be an effective part of the Internet. These rules define the formats of messages, how routing is done, conventions for managing the network elements, and so on.
APPENDIX A 250 Oversight of the Internet Internet Society The Internet's very decentralized organization is, of necessity, balanced by mechanisms for coordination and oversight of key issues and concerns. Of the several organizations that play a coordinating role, perhaps the most important is the Internet Society.23 The international Internet Society provides a means for global cooperation and coordination for the Internet and its internetworking technologies and applications. Its members reflect the breadth of the entire Internet community and include individuals, corporations, nonprofit organizations, and government agencies. The Internet Society's principal purpose is to maintain and extend the development and availability of the Internet and its associated technologies and applicationsâboth as an end in itself and as a means of enabling organizations, professions, and individuals worldwide to more effectively collaborate, cooperate, and innovate in their respective fields and interests. Its specific goals and purposes include the following: â¢ Development, maintenance, evolution, and dissemination of standards for the Internet and its internetworking technologies and applications. Internet standards are formulated by the Internet Engineering Task Force, an open-membership body that currently operates under the auspices of the Internet Society; â¢ Growth and evolution of the Internet architecture. The Internet Architecture Board (IAB), which offers architectural guidance for the Internet, sits under the Internet Society; â¢ Maintenance and evolution of effective administrative processes needed for operation of the global Internet and for internetworking. For example, the Internet Assigned Number Authority, which is responsible for the allocation of network numbers for the Internet, is a function of the IAB; â¢ Education and research related to the Internet and internetworking, and the collection and dissemination of relevant information; and â¢ Harmonization of actions and activities at international levels to facilitate the development and availability of the Internet. The Internet Society is currently establishing liaison relationships with such organizations as the International Organization for Standardization and the International Telecommunications Union. U.S. Federal Government As outlined above, the U.S. government historically has played a very
APPENDIX A 251 important role in the overall coordination of the Internet, but this role is diminishing with the Internet's increasing commercialization and internationalization. However, the government continues to provide some guidance in key areas, as in the recent award by NSF to provide coordination of Internet routing. In addition coordination among federal agency networking activities is provided through the Federal Networking Council (FNC; which was established in 1990 with links to the Office of Science and Technology Policy (OSTP) overall, the High Performance Computing and Communications Information Technology effort under OSTP, and the Office of Management and Budget) and the FNC Advisory Committee, which consists of representatives from the nongovernmental constituencies served by federal networks. The FNC oversees various working groups (e.g., the Engineering and Operations Working Group concerned with network technology design and implementation, and the Federal Engineering and Planning Group that conducts technical analysis and operational coordination of the federal agency networks).24 Paying for Use of the Internet In the early days of the Internet, most of the funding for the facilities and also for network use came from the U.S. government, as discussed in Chapter 5. Briefly, most of the costs today are covered by payment from the various users, both institutional and individual, that attach to and are a part of the Internet. A typical payment scheme is a monthly access fee, although other patterns of payment can be found. For an institution, the costs of installing, operating, and supporting the local networks and related infrastructure at the campus are usually much greater than the access fees paid to the regional or long-distance Internet provider. NOTES 1. Kleinrock, Leonard. 1976. "The Arpanet." Pp. 304-314 in Queueing Systems, Volume 2: Computer Applications. Section 5.4. Wiley Interscience, New York. 2. Cerf, Vinton. 1993. "How the Internet Came to Be." In The Online User's Encyclopedia by Bernard Aboba. Addison-Wesley, New York, November. 3. Cerf, Vinton. n.d. "A Brief History of the Internet and Related Networks," distributed electronically. 4. To facilitate communication and coordination among regional networks and coordination between those networks and NSF, the Federation of Academic Research Networks (FARNET) was founded independently in 1987. 5. Mandelbaum, Richard, and Paulette A. Mandelbaum. 1992. "The Strategic Future of the Mid- Level Networks." Pp. 59-118 in Building Information Infrastructure. Harvard University Press, Cambridge, Mass. 6. Aiken, Robert, Hans-Werner Braun, and Peter Ford. 1992. "NSF Implementation
APPENDIX A 252 Plan for the Interim NREN," GA-A21174, GA Project 3900, San Diego Supercomputer Center, draft, May. 7. Aiken et al., 1992, "NSF Implementation Plan for the Interim NREN," p. 8. 8. Aiken et al., 1992, "NSF Implementation Plan for the Interim NREN." 9. Stephen Wolff, National Science Foundation, remarks at National Net '94 conference, Washington, D.C., April 7. 10. "A key activity of the NREN program is to enhance the interconnection technologies and strategies for both federal and non-federal networks, without interfering with the autonomous management of each component network." Aiken et al., 1992, "NSF Implementation Plan for the Interim NREN," p. 3. 11. Computer Science and Technology Board (CTSB), National Research Council. 1988. Toward a National Research Network. National Academy Press, Washington, D.C. (The Computer Science and Technology Board became the Computer Science and Telecommunications Board in 1990.) 12. See agency matrix in Committee on Physical, Mathematical, and Engineering Sciences, Federal Coordinating Council for Science, Engineering, and Technology, Office of Science and Technology Policy. 1994. High Performance Computing and Communications: Toward a National Information Infrastructure. Office of Science and Technology Policy, Washington, D.C., pp. 26-27. 13. The Gigabit Testbed program is funded by NSF and ARPA and involves industry contributions and participation as well as participation by universities and government laboratories. See Committee on Physical, Mathematical, and Engineering Sciences, Federal Coordinating Council for Science, Engineering, and Technology, Office of Science and Technology Policy, 1994, High Performance Computing and Communications: Toward a National Information Infrastructure. 14. Cerf, 1993, "How the Internet Came to Be." 15. National Coordination Office for HPCC. 1994. Information Infrastructure Technology and Applications. Office of Science and Technology Policy, Washington, D.C., February. 16. For those parts of the internet installed by the NSF in support of scholarly and academic activities, there are policy restrictions on the uses of the network, but these restrictions are expected to vanish from the core of the network with the transition to commercial provision of Internet service. 17. Personal communication, Elise Gerich, Merit Inc., April 1994. 18. "[The routing arbiter's] role will be to promote Internet routing and stability, estabiish network topology and policy databases, develop procedures to resolve problems between network entities, develop advanced routing technologies, provide simplified routing strategies, and promote distributed operation and management of the Internet." Part of the challenge is to develop new technologies, including high-performance, workstation-based route servers. See University of Southern California. 1994. "NSF Announces Major Network Awards: California, New York, and Michigan Groups Win." Press Release, February 15, distributed electronically. 19. No firm number exists for the number of organizations involved; the range is perhaps from a lower bound of 2,000 to 3,000 to an upper bound of 10,000 to 12,000. Individuals contacted by the committee remarked on how interesting the question wasâit is a measure of the Internet's true decentralization that we cannot count or even identity all the Internet players. 20. Aiken et al., 1992, "NSF implementation Plan for the Interim NKEN"; and Huston, Geoff. 1993. "Connectivity Within the InternetâA Commentary." Australian Academic and Research Network, Canberra, Australia. 21. Stewart, Thomas A. 1994. "The Netplex: It's a New Silicon Valley," Fortune, March 7, pp. 98-104.
APPENDIX A 253 22. Lewis, Peter H. 1994. "A Traffic Jam on the Data Highway," New York Times, February 2, pp. DI and DS. 23. Material on the Internet Society was adapted from an electronic Frequently Asked Questions document, "What is the Internet Society?" dated March 5, 1994. 24. Aiken et al., 1992, "NSF Implementation Plan for the Interim NREN."