When the Net gets too hot for commerce to handle

Alistair Chalmers says Internet near meltdowns in the US and Australia offer lessons that the UK's academic network managers must learn

Most readers of this piece will be familiar with Janet, the United Kingdom's Joint Academic Network, and perhaps with its equivalents such as WiN (Germany), Renater (France), Surfnet (The Netherlands) and the many other national networks in higher education and the associated research community. For teaching and research and, increasingly, for administration, computer networking has moved from the status of a valued supplementary amenity to being a routine necessity.

With Janet, the UK has been in the van of networking for the past 15 years. It is only in recent months that the quality of SuperJanet, established four years ago, has begun to be rivalled anywhere outside the United States -and even there the comparable facilities are not as pervasive.

Some national networking budgets are pulling away from the UK's. For instance, we are pleased to have contracted for our first T3 transatlantic link. This will increase capacity from 21 to 53 megabits per second, on two links instead of nine. However, the Nordic countries have enjoyed such access for some time and Germany recently put not one but two 45Mbps links in place.

In Europe it remains fairly common for the government department or agency which carries responsibility for networking policy - usually the education or research ministry - to provide the entire budget. Where this is not the case, there are usually other arrangements which ensure that an adequate budget is sourced. These may be through multi-layered contributions - for instance, by the federal government and the Lander in Germany, by charges on the local institutions which are in practice met by national bodies (the National Foundation for Scientific Research in Portugal), by central government contributions to the costs of innovation and development (the Ministry of Education and Research in the Netherlands), by devolution of some operations and their costs to regional level (in France), and so on. Each country is in its own way juggling with various sources of funding to achieve substantial enhancements in provision.

In parallel, the global Internet (of which Janet is a component), its potential to become an "information superhighway", and its special applications - particularly the World Wide Web - is emerging as an enabling infrastructure with potential in every field, including commerce and entertainment. This trend has led to the assumption that the provision of networking, and of Internet access in particular, is rapidly becoming a commodity market.

Deregulation of telecoms provision is accelerating and is supposed to be complete within the European Union by 1998. Are we on the verge of an era of rapidly improving network provision at low prices? Will the requirements of higher education no longer be distinctive, demanding special efforts within the academic community to satisfy them? Will an organisation such as a university or a research council unit be able to obtain connectivity from a choice of suppliers?

Industry observers seem to agree that rumours of a low-price post-deregulation telecoms market in Europe are exaggerated. Compare, for instance, the slow fall in telecommunications prices in the deregulated UK with the markets for other information technology products and services. The main aim of most national governments is to privatise their state-owned monopolies for the highest price; and this requires that the cash-cow status of these operators be preserved at least up till the date of deregulation, and preferably beyond.

This aim is having a stronger effect than impending deregulation. The situation within the UK is one of the better ones. This, however, does not extend to the prices of international lines, which are about an order of magnitude greater than those for domestic links of similar length and are the main drain on the present budget.

Within the past two years, academic networking requirements in the US and Australia have been handed over, with only small residual constraints, to the mercies of the commercial marketplace. The experiences there are instructive.

Commercialisation of the Internet is well advanced in the US. Partly for this reason, two years ago the National Science Foundation Network Executive discontinued the NSFNet, the continent-wide academic network which was once the Internet's backbone. Client organisations were allocated an annual grant, tapering to zero over a period of five years, to assist with the purchase of an equivalent service from one of three public network operators. Clients generally bought their service through a regional consortium, and the network operators contracted to a number of conditions, including the free exchange of academic traffic. Within months, congestion had built up to unacceptable levels.

A year later, the board of the Australian Academic and Research Network transferred the AARNet staff to the state-owned public network operator, Telstra, and entered into a contract with Telstra for provision of the national network. The Australian topology is simpler than the UK's: AARNet interconnects one node on each of eight regional networks and provides the international services. This simplicity lends itself to a scheme of charging by the bit transferred, which gives the supplier an incentive to meet growth in demand. Even with these twin advantages of simple topology and usage-based charging, Australia encountered the same difficulties as the US. Telstra imposed significant price increases.

The AARNet board has completed a re-procurement of the national services, resulting in a contract with the recently established alternative national telecoms supplier, Optus. The board is reconsidering international provision. Some observers expect AARNet to resume direct management.

In the US, given that there is no longer a federal agency with a remit to provide general purpose networking, the response has been slower but more powerful, resulting in the Internet 2 project, whose goals are to provide a leading edge network for the national research community, to improve production Internet services for the academic community, and to enable a new generation of applications and services. The project is spearheaded by a consortium of national bodies with the support of regional organisations, individual universities, and commercial affiliates. Substantial public funding has been announced.

The US is able to buy its way out of trouble, spending not only money but also personal energies, and drawing on a huge reservoir of expertise. The Australians, like most of their European colleagues, do not have resources on that scale. They are able, however, to retrieve the situation because they did maintain a policy and planning function (the AARNet board) and retained a skeleton staff.

These two experiences suggest that any effective market in advanced network services is still some way off. In effect, from a situation in which there was widespread agreement that Internet services had become a commodity and that there was no need for special structures and funding for higher education and research ("We really thought it would work!"), our US colleagues have not only stared near-meltdown in the face but have also affirmed that Internet development is not a finite task but an upward spiral: as the services and applications of one generation of technology bed down and become commodities, the nature of teaching and research will be driving towards the next, using the community's own skills and experience. Internet 2, then, is not a one-off solution: it is likely to be a continuing, long-term strategy.

The conclusion is that the academic community must look to itself to ensure that its networking needs are met. We are still a long way from any situation in which market offerings might deliver all that is required. And central coordination is essential for forward planning, for effective cooperation, and for bulk purchasing, especially of expensive international capacity.

There are no good examples of this being achieved without a central budget and a special-purpose network administration.

The UK's present way of agreeing and creating a central budget is not the only possible one. Nor is our way of organising the administration. Many models are in operation in various European countries. However, without a budget and administration of some kind not only would the academic community's significant purchasing strength be dissipated but its distinctive requirements would scarcely be heard in the clamour of political and commercial lobbying for the so-called information superhighway. Higher education's networking requirements remain distinctive and continue to require foresight, planning, and collective effort.

Alistair Chalmers, honorary professor at the University of Sussex, completed his five-year term as chairman of the Advisory Committee on Networking on March 31. For part of that time he was also chairman of the European Consortium for Academic and Industrial Research Networking.