The United States would benefit significantly from the creation of a national research network (NRN). The National Research Network Review Committee has evaluated the proposal for an NRN submitted by the Office of Science and Technology Policy (OSTP), and the committee strongly supports the concept presented. An NRN would create a computer network infrastructure to provide much-needed support to the scientific research community. Data obtained by the committee regarding current and anticipated research activities demonstrate that an NRN could dramatically improve the productivity and quality of output of the U.S. research community. Through these direct .benefits, plus commercial spinoffs from associated computer and network research, an NRN could greatly promote U.S. competitiveness in a multiplicity of disciplines.
The OSTP proposals for an NRN have three phases, a structure the committee supports. In the near term (late 1980s), it would interconnect the existing networks that are fragmentary, overloaded, and poorly functioning. In the second phase (in the early 1990s), it would extend computer network support to researchers at more institutions than are currently served, in particular the smaller institutions not now served well, and it would significantly enhance the range and quality of network-based services provided to researchers. In the final phase (late 1990s) advanced technology and techniques would be deployed to provide a dramatic increase in speed and function to match the power of computers in use at that time.
The information-processing capability on the desk of most researchers today is awesome. The researcher’s workstation is an indispensable tool in his daily research endeavors. However, his productivity is being severely hampered by his inability to effectively link his high-performance workstation to the supercomputers, to the large scientific databases, and to his colleagues in a fashion that allows him to incorporate into his research these other resources. What is needed is an accessible, user-friendly, high-speed NRN to provide these connections. The science research community currently sees a serious mismatch between today’s high-performance processing engines and low-performance computer
networks. This situation is extremely unfortunate for the United States, the nation that spearheaded networking technology 20 years ago.
National research network users (the scientific research community) have two principal sets of concerns. First, they want—and the committee believes they should receive—an operational network that is stable and predictable, offers high-quality performance, and does not require advanced knowledge of networking technology to use it routinely or to cope with occasional problems. While the OSTP proposals emphasize goals for high speed, the committee equally ranks the concerns for widespread access, user-friendly service-orientation, quality, and performance. Second, users—especially researchers in disciplines that have been less computer- and network-intensive to date—are concerned that federal funding for a research network not diminish federal funding for the conduct of research. The committee shares this concern.
To encourage the development and use of a widely accessible network service, the committee suggests that the NRN might provide (1) a universal basic service at some low to moderate speed for electronic mail at a low cost to users and (2) higher levels of service with an appropriate charging structure. A low charge for basic service would lessen the initial resistance that might otherwise be felt by first-time users and yet discourage the overloading that occurs with free networks.
There will always be a tradeoff between science research users’ needs for a stable operational network and network researchers’ desires for high performance using the latest networking technology. This latter desire arises both from the developers of network technology, who need a testbed for new ideas, and from that sophisticated class of users with advanced requirements. Because network technology is evolving, there will always be a current networking practice, developing network technologies, and research targets. The challenge underscored by the committee is to manage the progressive introduction of increasingly sophisticated technology without compromising users’ ability to conduct the research for which the network is a tool. This may not be possible without substantial investment, and, even with it, compromises may be necessary.
The committee viewed the OSTP proposals as providing a general direction for the NRN, not a set of design specifications (which they were not) to be evaluated. Four areas of concern were addressed: technical considerations, funding, management, and commercial and specialty networks. The recommendations are summarized below:
Technical considerations. The technical demands for the first two phases identified by OSTP are not leading edge items. Indeed, during these phases the NRN will be catching up with the state-of-the-art technology. Phase 1 requires coordination; phase 2 requires careful attention to network management. However, achieving the high-performance network carrying the high volumes of traffic envisioned in phase 3 for the late 1990s requires computer and communications
research using a clean-sheet approach that must begin now. The phase 3 network is dominated by technical issues, in particular those associated with network control but also including appropriate technology for switching, routing, multiplexing, processor interfaces, protocols, connection-oriented communications, and layered architectures. All of these areas offer opportunities for technology transfer to industry from the necessary research program.
Funding. There are a number of critical challenges associated with funding for an NRN. These challenges must be recognized and addressed up front, and the transition to a new research funding regime must absolutely lead to a stable funding mechanism or the NRN will never achieve the impact of which it is capable. There are two coupled issues here: (1) funding for network technology and (2) charging for network use. The committee recommends that economic incentives be provided to encourage industry to provide existing network technology to the NRN, but that the government provide funding for advanced network development in cooperation with industrial research efforts where possible. These funds for advanced development should be kept separate from funds allocated to running the operational network. A mechanism for providing (earmarked) funds to users to defray user charges for network use must be established. This mechanism should allow the user to choose among network services. An example of such a mechanism is the use of a voucher for network services.
This transition to user charges requires sensitivity to research users’ concerns. For example, one challenge is to introduce charges for NRN use to researchers who may be accustomed to “free” service from such networks as Arpanet. User charges would allocate resources and inhibit excessive demand for use of finite network facilities. A charging structure would also make possible an eventual transition of the service to the private sector. While the economic principles are obvious, the committee recognizes that introducing charges will be difficult and will require compensatory changes in federal grant practices, such as providing for (additional) computer networking costs in research grants.
Management. The scope and scale of the NRN create an imperative for careful attention to key management issues. Good management is absolutely essential for maximizing the effectiveness of the investment in an NRN. It is a critical issue that requires early (and continuing) attention and action, and it is a nontrivial element of the total cost of providing network service. The overall problems are to engineer facilities and services to match demand and tune them to the expressed needs of users, to maintain them in an operational state, and to operate them economically. Network facilities management, operational service management, and process management (for example, introduction and phasing-in of
user network services and facilities) each require commitment and resources now. The urgency of the management imperative is underscored by the committee, which points to inadequate service in existing research networks (notably Arpanet and the preexpansion NSFNET) and to the challenges implied by extending access to what could eventually be almost a half-million users in the research community. Poor management and the absence of user services are potential barriers that could prevent the NRN from realizing its potential. To overcome these barriers, an operating structure that does not now exist must be created. It should consist of a funding mechanism through normal government channels, an oversight committee, and an experienced research executive from the private sector for day-to-day management; in addition, one or more user groups should be established to provide input from and output to the user community.
Commercial and Specialty Networks. Discussions of an NRN can easily assume a stand-alone network, but the committee concluded that the proposed network must build on existing and anticipated commercial network facilities and network-based services. It encourages those responsible for developing and implementing the NRN to work with the common carriers and computer vendors; it also encourages long-range planning that would allow for the eventual transition of at least portions of a government-developed NRN to industry, because meeting the service and access goals of the network implies resources beyond the apparent capabilities of federal mission agencies or the National Science Foundation. The commercial network suppliers have decades of experience in network research and in providing accessible and reliable service to millions of users; it would be prudent to leverage this experience for the NRN.
We have a special opportunity with the NRN. We can give an enormous boost to U.S. competitiveness by providing this networking infrastructure. But we must not hesitate, for our foreign competitors are busy at work with their own national research networks (e.g., the RACE program in Europe and national network projects in Canada, Japan, and in newly industrializing countries such as Singapore, and so on). If we are successful in the NRN development, our competitive advantage will continue to grow as this unique infrastructure dramatically improves the quality of output and productivity of the research community of this nation.