History of Research Networking
The history of research networking is a surprisingly short one. Although automatic telecommunications networks have been around for more than a century and computers have been available for nearly fifty years, serious attempts to combine the two technologies only really started in the mid 1970s. Research networking is primarily concerned with the linking of computers and communications. Initiatives of research networkers both in Europe and North America in this period were fundamental for the creation of the data networking environment that we take for granted today. The development, service and commercial cultures of the computer industry and the telecommunications industry have always been quite different. Research networking has bridged and continues to bridge the gap between them.
The first research networks in Europe started to emerge as national initiatives. They concentrated on connecting major university and research sites together. The transmission speeds then were, by today’s standards, remarkably low. The telecommunications networks which were used at that time were starting to make the transition from analogue to digital technology. They were entirely optimised for voice telephony, which was the dominant traffic type. In Europe, the telecommunications industry was organised around a set of national monopoly organisations, typically combined with national postal organisations, who had absolute monopolies of nearly all aspects of service provision. This frequently included even “in building” wiring. These monopolies were technically conservative, lacked innovative forces and typically viewed data communications with considerable suspicion. They saw data as a service that computer companies, especially in those days IBM, would use to try and undermine their lucrative equipment and service monopolies. A broadly similar picture could be observed in the USA. In the USA however, the monopolies were more limited. Government sponsored “Computer Inquiries” which addressed the issues of computer communications led, in 1982, to the break up of the nation-wide Bell system and to the emergence of a more competitive national market.
International telecommunications remained a jealously guarded monopoly world-wide. International voice services, with their complex bilateral transfer charging arrangements between sovereign national operators, represented a major “cash cow” which governments were reluctant to kill. This was a particular problem in Europe as the set of national markets were, individually, much smaller than the US market. The introduction of digital switches and digital transmission in telecommunications during the 1980s, replacing mechanical switches and analogue transmission technology, meant that the basic technology of the telecommunications industry became much more suitable for data communications. Access to this technology was, however, limited internationally as it would enable alternative network providers to bypass the transfer pricing regimes for international voice services that were so profitable for the monopoly operators. Again, this was a particular problem in Europe. Mainly as a consequence of this, the development of international research networking In Europe lagged seriously behind national initiatives, especially those in the United States.
The Early Days of European Research Networking
The liberalisation of European telecommunications markets was a gradual process. There was always some very limited competition in regional markets within the Nordic region, but, for most of Europe, telecommunications was managed by the Postal Monopoly. The first changes came when the management of telecommunications was separated from that of the postal service. This did little for competition, but at least focussed the businesses more on the provision of service. The first really competitive national telecommunications market was in the UK. By 1983, there was a real competitive market for telecommunications. As a result, the UK academic network (JANET) was one of the early innovators in Research Networking. Other countries, seeing the benefits of liberalisation in the UK followed suit. The national markets for telecommunications gradually became more competitive and National Research networks took advantage of this liberalisation.
The pan-European Dimension
Developments on a pan-European scale were much slower. This was partly due to the lack of a common organisation but also to the fact that the market for international telecommunications in Europe was limited and monopolistic. The national research networks got together in 1986 to found RARE (Réseaux Associés pour la Recherche Européenne). In 1987 a number of European countries, co-operatively with the European Commission established the COSINE initiative. Cosine had two objectives. The first objective was to establish a common pan-European infrastructure to connect the National Research and Education networks together. The second objective was to foster the implementation of Open Systems Interconnection hence the name COSINE (Co-operation for Open Systems Interconnection in Europe). The overall plan for COSINE was very ambitious. It consisted of the construction of a pan-European X25 network to interconnect National Research networks, together with a programme of OSI related developments, covering a broad range of areas from file transfer protocols (FTP), through the provision of gateways to OSI services, and the provision of information services to targetted user support. RARE was the contractor for this work and set up a separate Project Management Unit to carry out the work. Unfortunately, there was a significant time lag between the establishment of the goals of the COSINE project and the start of its implementation. In this time, the Internet, router-based networking technology and the associated suite of applications protocols started to have significant impact. The applications protocols were vendor independent, thus fulfilling one of the major goals of OSI. They were also, in general, simpler than the OSI protocols and had more practical implementations. The Internet networking technology, whilst less manageable than X25 was, however, much more ubiquitous. X25 implementations in Europe suffered badly from being country specific, thus portability of terminals and applications was a serious problem. In contrast IP was, relatively speaking, a “plug and play” technology.
The initial network built by the COSINE project, IXI (International X25 Infrastructure), achieved one of the goals of the project, namely to create a pan-European Infrastructure to interconnect the National Research Networks in Europe. It connected 18 countries together using for the most part 64Kbps leased circuits. It was, however not efficient at meeting the emerging demand for IP services. It was succeeded, in 1992, by the EuropaNET network which had the ability to support both X25 and IP protocols. It bridged the gap between the two potential networking solutions and allowed a migration from X25 to IP. EuropaNET was the principle achievement of the COSINE project. Whilst much of the OSI development work was abandoned, as users switched to the Internet protocol suite, EuropaNET represented the networking platform upon which successive generations of pan-European network have been built. EuropaNET was based on 2Mbps leased circuits. In comparison to IXI this was a major improvement. The COSINE project also achieved another result. It was responsible for creating DANTE.
COSINE had created an infrastructure, and it was recognised by those who were responsible for funding the project that if this infrastructure was to survive and develop, it was necessary to have a permanent organisation to continue to support and develop the network. As part of the project a working group was established to make proposals for the future organisation of European Research networking. This group produced a blueprint for the future organisation which was described in the report “Towards a Single European Infrastructure” .
The DANTE Era
DANTE today is very much the implementation of this blueprint. DANTE, which actually stands for Delivery of Advanced Networking Technology to Europe, was established in 1993 in Cambridge, UK as a limited liability company. The company is a “Not for Profit” company and has a special tax exempt status that has been granted by the UK government. The location of Cambridge was chosen as a result of an international competition. The tax benefits which the UK government offered were a major factor in determining the choice of location.
EuropaNET represented the second generation of pan-European Research network and existed for three years. It was succeeded by a new network, EuropaNET 2, which was broadly similar to the first EuropaNET network, with one significant development: EuropaNET 2 included direct global connections between Europe and research networks in the USA and Japan. At that time, global research connectivity was generally organised on a national basis. The global connectivity provided by EuropaNET 2 was typically used by the smaller European NREN’s or those in geographically disadvantaged locations where non-European connectivity was either unavailable or very expensive. EuropaNET 2 was relatively short-lived, being superceded in 1996 by a new network, TEN-34, that was equally short-lived. TEN-34 was the first pan-European network that exploited some of the early benefits of liberalisation. TEN-34 was assembled from a number of components that became available as new network providers started to try and exploit the liberalising environment. An element of TEN-34 was provided by Unisource (a new network operator that has subsequently failed) who wished to find a trial customer for the pan-European IP service that they were seeking to launch between Sweden, the Netherlands, Germany, the UK and Spain. This was complemented by ATM connectivity provided between the traditional telcos of France, Germany, the UK and Italy. All in all TEN-34 was less than optimal in terms of network development. It combined a variety of elements that were available at the time in a pragmatic development to improve connectivity among European research networks. It was, however, innovative in two respects. Firstly, it combined differing elements from competing suppliers. Secondly, DANTE was responsible for integrating the elements together and providing a network management service to ensure that the assembled components were operated as a single network.
TEN-34 was succeeded in 1998 by TEN-155. The TEN-155 network represented a significant improvement, in terms of performance, architecture, and service portfolio. It was the first pan-European network really to benefit from the liberalisation of the European telecommunications market. The availability of 155Mbps links, as a key building block, at a fraction of the price of previous connectivity, was the first real sign of a more competitive environment. Technically, TEN-155 combined ATM and IP technology in an innovative solution that allowed the creation of virtual private networks as well as guaranteed performance. TEN-155 existed for three years. During its lifetime key elements of the network were upgraded to 622Mbps, thus rendering the eponymous network name somewhat redundant. It did however illustrate the limitations of ATM technology as this technology was unable to cope with the upgrade to 622Mbps. For the later stages of its life, TEN-155 was a mixture of pure IP service and IP over ATM service.
GÉANT took over from TEN-155 in December 2001. GÉANT was the first generation of pan-European network that can reasonably be described as world-leading. It combined 10Gbps DWDM connections with high performance routers over a geographic scale that had never been achieved previously. The reduction in costs facilitated by market liberalisation, combined with access to the most advanced transmission and routing technology, enabled DANTE to build a network which had over 40 times the network capacity of its predecessor network. In addition, GEANT integrated global research connectivity between Europe, North America and Japan as the basis for truly global research network co-operation.
GÉANT2, which began in September 2004, represented yet another step-change in pan-European research networking. The hybrid network, comprising both routed IP and switched components, has been designed to provide the greatest possible degree of flexibility to Europe's researchers. In particular, it seeks to address the increasingly common scenario of researchers with very specialised and precise networking requirements, who need high bandwidth and stringent quality of service measures to support their work. Multiple 10Gbps wavelengths have been deployed at the network's core, and the overall capacity of the network is enormously increased by comparison with its predecessor.
GÉANT is the third generation of the GÉANT network, and successor to GÉANT2 which began in April 2009. It continues and develops the GÉANT network and builds on the technology and service elements of GÉANT2.
