The Evolving National Information Network
As the international networking environment continues to develop, the technology trends already established seem likely to continue. Protocols. are continuing to develop and converge, with a steadily increasing commercial activity growing up to develop and sell the products required to implement and use the standards. The use of the TCP/IP protocols for intracompany networking is accelerating rapidly. Network speeds are increasing. For instance, the NSFNET backbone put its first 45 megabit circuits in service before the end of 1990, and is now providing 45 megabit service throughout the network. A grant to the Corporation for National Research Initiatives from DARPA and NSF is supporting a number of testbed projects through which industry and higher education are building and experimenting with networks capable of transmitting information at speeds greater than 500 megabits. At the same timeS the success of the NSFNET effort has generated substantial interest on the part of the major telecommunications vendors, many of which are carrying out technology development efforts in their corporate research and development laboratories.
Experiments are also underway to further develop new uses. In addition to the client-server and distributed computing environment work referred to above, experiments are now underway on the transmission of Group IV facsimile over the internet. Demonstrations have also shown the feasibility of transmitting video and voice on the packet network, although further standards work needs to be done before high quality can be provided over congested networks.
The evolving international packet data infrastructure is organizationally both highly non-traditional and extremely complex. It may in fact be an exemplar of organizations for our future, but its very novel structure generates interesting issues for both the supporters of the network and the traditional organizations with which the network must interact. First of all, the network is chiefly held together by a set of shared protocols and values rather than hierarchical management structures and contracts. The protocols and values support a high level of collaborative activity, some of which is governed by legal contracts, but most of which has been carried out in an informal, highly flexible fashion.
It is instructive to trace the path of a packet originating in my home computer in Ann Arbor, Michigan which is bound for a colleague’s computer in Ireland. The packet from my computer first enters an Ethernet connecting several different computers in my home. I manage the TCP/IP network connecting those computers as a private network. One of the computers serves as a gateway to an ISDN line provided by Michigan Bell Telephone through a cooperative development project with the University of Michigan. When the packet arrives from the Ethernet to the gateway computer, the address it carries informs the computer that it is next bound for the University of Michigan’s campus data network. As a result, the computer signals Michigan Bell Telephone’s Ann Arbor central office switch to set up a digital connection from my computer over the ISDN line to a corresponding gateway computer in the University of Michigan’s Computing Center. After the Michigan Bell switch responds and establishes the connection, my gateway computer delivers the packet to the University of Michigan gateway computer. That computer, also connected to an Ethernet, sees the address and routes the packet through the Ethernet to the University of Michigan backbone network. From there, the packet travels to the statewide MichNet, which in turn passes it on to the NSFNET backbone network. Once on the NSFNET backbone, the packet travels to a gateway computer in New York which places it on an undersea fiber optic link running to Switzerland where the packets are received by a European supercomputing network. The packet is transmitted over that network via Paris and London to Ireland. Ultimately, the packet reaches a computer on the campus in Dublin, whereupon it is placed on a campus network and delivered to the machine being used by my Irish colleague. Amazingly, all of this activity typically takes place in about a second.
In this process, no fewer than fifteen computers have been involved, owned and operated by at least 10 separate organizations, none of which are under any direct hierarchical management control and most of which can be said to coordinate their activities only in the loosest possible fashion. Yet, the reliability of information delivery through this complicated international web of organizations is in general higher than placing a telephone call or sending a paper letter to the same place. The shared protocols enable this operation, and the shared values of the individuals in all of the participating organizations are responsible for the reliable operation of this complex system.
Not only is this collection of organizations extensive, it is also characterized by agreements ranging from research and development partnerships to contracts for service and not a few handshake agreements. Most of the organizations now providing these linkages did not exist as little as five years ago. One of the significant issues in the planning and further development of this worldwide packet network infrastructure is whether it should continue to develop as a loose confederation of many different types of organizations or whether there should be an effort made to create more of a traditional hierarchical organization controlling the activities. As different types of organizations with varying objectives provide service in support of the network or attach to the network to allow use, the nature of the required arrangements has become steadily more complex.
For instance, when all of the users of the network were associated with a research organization or an institution of higher education, it was straightforward to make an argument that the infrastructure merited support from federal and state government. Now that the users have a broader range of relationships and the network includes for profit activities such as information provision, it has become less and less clear which part of the infrastructure should be supported by government and which part should be supported by user fees. Within academic organizations, when the use of the facility was primarily the support of research, it was appropriate to think of the financial support coming from overhead on or direct costs of research activities. As the uses expand to include instruction, library access, and facilitation of the operations of the institution, it is less clear that the infrastructure should be supported research funds. In the current financially constrained environment for higher education, various institutions have made differing ad hoc arrangements for network funding.
In this environment, new organizations are being created to fill in the gaps between the commercial and the not-for-profit community and to provide an avenue for increased funding of the network infrastructure through user fees. Among those recently established are AlterNet, Performance Systems International (PSI), and Advanced Network and Services (ANS), all of which offer TCP/IP packet transport connections and service via facilities they own and connections to the internet. In addition the major telecommunication carriers are beginning to enter the TCP/IP packet transport market, with recent initiatives from Sprint, AT&T, and MCI.
Support from the federal government is also steadily growing. The High Performance Computing and Communications (HPPC) program authorized in 1992 included over $90 million of funding for the development of the national network, all directed at creating a National Research and Education Network (NREN). The HPPC initiative involves the Department of Defense, the Department of Energy, the National Aeronautics and Space Administration, the National Science Foundation, the Department of Commerce, the Environmental Protection Agency, and the National Institutes of Health. The Clinton administration is making a major effort to more rapidly develop the National Information Infrastructure, and the NREN is clearly linked to that effort.