The Evolving National Information Network
What We Have Learned
Requirements
All of this experience in nationally accessible packet switched networking has verified three key requirements for the usefulness of the network.
Very wide range of network capacity for each user
First, use of the network by any individual user almost inevitably occurs in bursts of activity. Much of the time the network connection is idle or carrying a very low rate of traffic in support of remote logon to a computer running a character by character application. From time to time, however, such as when a file needs to be moved from one computer to another, or a package of electronic mail needs to be conveyed from one location to another, there is a need to move a comparatively large amount of information in a relatively short time. A researcher may need to convenient and simultaneously remotely logon to a computer supporting remote visualization o supercomputer results, access a bibliographic database, transmit a revisable document for remote editing, and convey and print the content of an article. As we move into client server models and ultimately to the distributed computing environment, this unevenness of network utilization will increase still further. As s stem evolve toward keeping the most-used information close to the user, information will flow over the network in ways not directly related to the user’s activity at the workstation. The network supporting this activity must have the instantaneous capacity to run at very high speeds when necessary while incurring minimum costs when demand is very small or non-existent. Packet switching networks are ideal for this sort of use.
Broad addressability reaching user desktop
Second, we have discovered that.t scholars and researchers need to have nearly simultaneous access to a very broad array of resources. At the same time a researcher is composing electronic mail, (s)he is likely to need to retrieve some information from a database, monitor a large computation going on a remote computer, nd have access to a computer bulletin board, all at different network addresses. In order to accomplish this, it is important that the user be able to simultaneously attach to multiple resources through the network. This capability {)f reaching out for multiple resources enables the user to customize a local computing environment to meet unique needs.
Ties to international industrial and scholarly community
Third, it has become clear that today’s researcher or scholar works with an international set of colleagues who themselves are affiliated with organizations not only in higher education but also in industry and government. The network infrastructure needs to provide connections not only within any of these communities, but across these communities. A scholar needs access to commercial databases just as much as (s)he needs access to his or her university’s library. A researcher needs access to the supercomputer and data banks at a national laboratory just as much as (s)he needs access to his or her own department’s minicomputer. A product engineer needs access to his or her research and development colleagues in the university just as much as (s)he needs access to the parts database upon which his or her product is being built. As a result, the network, while it may require a multiplicity of business and financial arrangements, should provide transparent connectivity across all of the sectors of the economy and providers of information.
Rapid growth underway
Our early experience has also made clear that, even in its comparative infancy, network access is a highly desirable facility. Every standards-based, packet switched network initiative has grown rapidly in the number of computers and users connected and in the amount of traffic it carries. For instance, traffic on the NSFNET backbone has grown more than ten percent per month over the last five years. In fact, the research and scholarly community uses information intensively and extensively and depends heavily on communication with one another to accomplish results.
The need for a broadly cooperative effort
The way in which standards facilitate very broadly distributed investment in support of the creation of a common service has made it possible for funds from an extraordinary variety of sources to be focused OIl the creation and expansion of the network environment. NSFNET is funded by a combination of educational, industry, and government grants and fees, with the great majority of funding coming from education and industry. Funding from the National Science Foundation is crucial to provide the facility which acts as a stimulus for broad funding, but represents less than ten percent of the total funding of this important resource. On the basis of the experience to date, it is clear that the education and research community is willing to continue to invest substantial resources in expanding both the capacity and breadth of access of the national networking environment.
The future
The higher education and research community is increasingly assuming the NREN’s existence built on the experience provided by the NSFNET. There remain, however, a number of difficult policy, funding, and organizational issues.
