Projects reviewed in this article:
Building the networks of the future has become a global business of particular concern to educational institutions. In the U.S., the educational use of the Internet is entering its second phase with Internet II, a backbone network designed to transport data between colleges and universities at high speeds. It was proposed by college officials and is supported by the Clinton administration. (Nearly 100 colleges and universities signed on as charter members of Internet II.)
But the planned changes in the infrastructure alone won't bring the next generation of networks. The challenge of the future is to extend the capabilities of the infrastructure to the user's desktop. While the infrastructure seems to be in good shape, many institutions are still struggling to meet the requirements of Internet II.
The foundation for the development of the information superhighway's infrastructure was laid more than a decade ago. The funding for Supercomputing Centers was established in 1984; in 1993, the National Science Foundation (NSF--http://www.nsf.org) decided to concentrate on creating a very high speed Backbone Network Service (vBNS), connecting the Supercomputing Centers and Network Access Points at speeds of up to 2.5 gigabits (2.5 billion bits) per second. As the initial users of the vBNS, the NSF-supported Supercomputing Centers (The Cornell Theory Center, The Pittsburgh Supercomputing Center, The National Center for Supercomputing Applications, The San Diego Supercomputer Center and the National Center for Atmospheric Research) have been experimenting with the enhancement of networks. (There currently is some debate as to what role the vBNS will play in Internet II. While many see it as a part of Internet II, others want to make the systems totally separate.)
The NSFnet also provided other high-speed links for institutions that foster new high-bandwidth applications. That network has been privatized (the frequent congestion of the commercial replacement has often caused complaints), and the program based on supercomputing "centers" will be replaced by a Partnership for Advanced Computational Infrastructure (see IA Vol. 1 No.1: "Will the Teraflop Flop?).
Through a collaborative effort of industry, academia, and government, with funding from the NSF, the Advanced Research Projects Agency (ARPA), and industry, five U.S. Gigabit Testbeds were created from 1990 through 1995. The project was coordinated by the Corporation for National Research Initiatives (CNRI), a non-profit research and development organization formed in 1986 with the goal to develop U.S. information processing technology and engage in scientific research on the design of experimental infrastructure.
The five original testbeds are known as Aurora, Blanca, Casa, Nectar, and Vistanet and since their establishment various other testbeds, e.g. the Magic wide area testbed, have been developed. In the San Francisco Bay Area, fourteen organizations have been working with the Pacific Bell Broadband Strategy and Product Development Group to develop and deploy a large-scale metropolitan-area network based on Asynchronous Transfer Mode (ATM)--a method of transmitting data widely used with fiber-optic cables. BAGNet (Bay Area Gigabit Testbed, http://www-itg.lbl.gov/BAGNet.html ) was built to investigate the computer multimedia network infrastructure needed to support a diverse set of applications.
In a joint effort, various European countries are also upgrading existing national networks and building ATM broadband networks to bring the European network infrastructure up to the standard currently in use or under development in the U.S.
Broadband networks allowing communications at 34-155 megabits have been developed in the UK (SuperJANET), the Netherlands, France and Norway (Supernet). Networks upgrades are also underway in Germany, Finland, Austria and Sweden. At a pan-European level, an agreement to develop a pilot ATM network in Europe (EuropaNet) was signed by six European network operators in November 1992 and now has 18 signatories.
While the future of the infrastructure looks promising, the future of local network connections remains uncertain. In reality, most campus networks don't meet the needs of users and researchers. The quality of multimedia, full-motion video, and live audio broadcasts depends on reliable, fast, high-quality connections from office computers to the network. Therefore the connections of office computers to the network have to be upgraded--with new circuit boards, switches and better cables--to enable users to take full advantage of the changes in the infrastructure. Right now, the majority of users on campus has an Ethernet connection with a top speed of 10 megabits per second. If the connection is used in a "dedicated" configuration, it will provide individual users with the maximum speed; in a "shared" configuration, an Ethernet line is hared by several people and the network speed is affected by the number of users.
Administrators are in the unenviable position that they have to make choices about the technologies their campuses will use to solve existing problems, while nobody seems to know what technologies will best be suited for campus needs in a few years from now.
Case Western Reserve University, one of the institutions taking part in the Internet II project, has been able to offer individual users connections that are as fast as 155 megabits per second by basing the campus network on Asynchronous Transfer Mode. Case Western's approach is unusual, since ATM is generally used for moving data across major network backbones. To take advantage of the technology and deliver speeds of 155 megabits per second, the computers have to be connected directly into fiber-optic cables, which normally make up the backbone that runs across a campus, or are used to connect the backbone to buildings. Most of the networks within buildings rely on the cheaper copper wires, some of which can carry data as fast as 100 megabits per second. Every desktop connected to an ATM network needs a special "card" or circuit board, which can cost $ 1,000 or more (although prices are dropping rapidly, and partnerships with technology companies might help to lower costs). Using ATM for the entire network, including individual connections, may yet turn out to be problematic, since the technology is still new. Campus technicians may have to be retrained to work with ATM cards, and bugs may have to be fixed.
Many educational institutions hope that faster Ethernet technologies will be a short-term solution to the problem. Fast Ethernet, which offers speeds of 100 megabits per second, is being considered as a viable alternative to ATM. Ethernet cards (standard or fast) cost one-tenth of an ATM card. The next step will be Gigabit Ethernet, which is 10 times faster than Fast Ethernet. The Gigabit Ethernet Alliance (http://www.gigabit-ethernet.org/) is currently busy promoting industry cooperation in the development of Gigabit Ethernet. The primary goal of the alliance is to support the work on Gigabit Ethernet standards that is being conducted by a Task Group (IEEE 802.3) and to contribute technical resources. The Group has already finalized the core proposals that will form the basis for the first draft of the Gigabit Ethernet standard; no major new technical proposals will be adopted. The core proposals define the scope of the changes required to adapt Ethernet for operation at 1000 megabits. The Task Group will meet again from January 27-29, 1997, to review the first draft of the standard, and by July this year, the specification may be complete and ready for the ballot process.
Sophisticated technologies continue to be under development, prices on hardware are dropping quickly, and there seems to be no clear idea what will work best with Internet II. For many educational institutions, network decisions still require an educated guess.
|
