Network (JANET)
3.7 Hindsight
Manchester universities, and the SERC’s Rutherford and Daresbury laborato- ries. Much of the additional funding came as a result of proposals for major upgrading of the computing facilities at these four sites, with a consequent need for improved network access particularly for interactive work. The technical situation was also improving, in two senses.
The formation of British Telecom (BT), had led to improved availability of higher data rate connections. This reflected a change in the way these links were supplied. Until this time, the links were provided by using a modem to connect to what was essentially a speech (analogue) circuit. Now, BT were prepared to allow direct access to the digital circuits which had were used to carry traffic between its exchanges. This gave the public access to digital “Kilostream” and
“Megastream” circuits with bit rates up to 64 Kbps and 2 Mbps respectively, much higher than the 9.6 Kbps and 48 Kbps with which JANET started. BT also had equipment which allowed a single Megastream circuit to be treated as a number of separately configurable 64 Kbps circuits, and by using this the NE were able to create what was effectively a fully linked configuration between the four major switches at London, Manchester, Rutherford and Daresbury. In par- allel with this, GEC were able to offer higher speed units to attach to their switches, which allowed these higher line speeds to be exploited.
Welcome as they were, the improvements in hardware were accompanied by an even more welcome relaxation in the attitude towards protocols. By 1986, it was becomingly increasingly clear that the connection oriented coloured book pro- tocols were not competitive with the connectionless ARPA protocols. It is an open question whether this represented an inherent superiority in the connectionless protocols, better implementations of them, or their availability on better hardware. Whatever the reason, the UK academic community had experimented with various means of allowing the two protocol stacks to coex- ist. These experiments had shown that this was technically feasible, and for some five years it was hoped that JANET could operate permanently in this way.
It was not until 1991 that use of the coloured books was finally phased out.
technically civil servants employed by a research council was by this time creat- ing more problems than it solved. Accordingly, David Hartley took steps to cre- ate the United Kingdom Education and Research Network Association (UKERNA), thus removing it from the necessity to conform to civil service norms for recruitment and promotion. UKERNA was, and to this day still is, a not for profit body, which is responsible for all aspects of running JANET, under contract to the Joint Information Systems Committee (JISC), which replaced the Computer Board in 1991.
Another of JANET’s strengths was that it was essentially a bearer network.
Applications programs were, and still are, almost invariably developed by end- users, who are of course in the best position to judge their own needs. The policy was, and still is, to develop only those products and services which are inten- sively network oriented, and can reasonably seen as a central meeting of a wide- spread demand; examples are the formation of a unit to monitor JANET performance, and the provision of Video Conferencing services.
JANET started in 1984 serving about fifty Universities, and about ten research council sites. The exact number of end-users was not known. At that time, most universities limited access to their computing resources to staff and graduate stu- dents, usually in science or engineering, and to a small number of undergradu- ates these same disciplines. Most SERC and NERC staff were computer users. I estimate that in 1984 there were about thirty thousand end-users of JANET. With the conversion of Polytechnics into universities, and the creation of many new universities, the number of sites which have “as of right”, and centrally funded, access to JANET has increased hugely. In addition, many other sites are “associ- ates”, with either a direct connection to a JANET switch, or a connection to an “as of right” site which provides onwards connection to JANET. The total number of end-users is still not known, but is thought to be about eighteen million, or 30%
of the UK’s population! When JANET started, the links between switches ran at 48 Kbps, and the links into universities or research councils at 9,600 bps. Today, the central core of JANET, or SuperJANET, operates at 10 Gbps (a Gigabit is one thou- sand million bits) between switches, and a typical site connects at 625 Mbps.
Details of the current configuration are available on the JANET web site at http://www.ja.net/. These are impressive figures for growth in scale and improvements in performance. The performance improvements come largely from improved technology, but it is important to understand that JANET has always been at the forefront in exploiting the best products, and has often been a pioneer customer. The growth in size reflects the confidence in JANET of the existing user base at any one time, and the willingness of others in the academic sector to participate. I think the Particle Physics community, and the Library sec- tor have both been especially effective in this respect.
With hindsight, we should have been much more aggressive in resisting the pressures to use the CCITT rather than the ARPA protocols, and to follow the procurement policy which restricted us to buying only UK products. I am rea- sonably sure that if at the very outset we had tried to overrule the procurement policy, we would not have been able to create JANET. I believe that had I fought harder, earlier we would have been able to migrate sooner.
There is one area in which I am sure that JANET has been hugely successful, that of breaking a big task down into a lot of small ones, and returning most of these tasks to the end user community. The JNT has always largely restricted its role to that of negotiator, bringing together interested would-be users of a new type of service with would-be suppliers, and then letting nature takes it course. Only in those rare cases where the realization of the new service implies alterations to JANET as a bearer network has there been central intervention. This has allowed the central role of JANET, as the bearer network, to be carefully insu- lated from the successes or failures of the hundreds of projects which JANET supports.
I would like to finish by giving my thanks to all those who have worked to make JANET the success it has undoubtedly been, and continues to be.
References
Flowers, B. (1966)University Grants Committee, Report of a Joint Working Party on Computers for Research, Cmnd. 2883. HMSO, London.
Verdon, F. P. and Wells, M. (1995) Computing in British Universities: The Computer Board, 1966–1991.The Computer Journal,38(10), 822–30.
JANET web site:http://www.ja.net/