The European Researchers’ Network Francois Fluckiger CERN, Geneva

Published in La Recherche, Issue number 328, February 2000 “Special , l’Avenir du Web”

In the middle of the 1980s, most European countries did not dispose of any operational network. Whilst Governments (States) and the major computer manufacturers negotiate, public research computer scientists adopt the Internet technology. Without explicit mandates from their organisations, they deploy it, spreading it out all over, and ten years later, the door is opened to the private sector.

François In spite of what has often science resources? In the United States, Fluckiger been said, the Internet is not the Arpanet network, set up in 1969, is, at the time of this a positive product of links universities and laboratories that article is published, in unconstrained liberalism, collaborate in research projects of the charge of the CERN nor a product of the Department of Defense. (See article by Web Services in imagination of private Jean-Claude Guédon p. 16.) But the Geneva. He is also enterprise which is free of technology used at that time to transport adjunct professor at the the straightjacket of data is not the one we know today – University of Geneva, member of the Advisory autocratic policy; neither “TCP/IP” technology will be introduced Council of the Internet could have accomplished into Arpanet only in 1983. What can Society (ISOC) and something that both researchers do with this network? expert for the European Governments and major Exchange electronic messages, transfer Commission. institutions had failed to files from one computer to another and achieve. connect at long distance to another No, the setting up of a world-wide computer. Moreover, the USENet information network, based on universal network, since 1979, allows any standards, is truly the result of public university who so wishes to exchange research: the Internet is the indirect any news item, the famous Internet product of universities and laboratories news, between any computer that has that not only invented the technology the Unix operating system installed, but – a rare fact – used it as a world- thanks to a technique developed by Bell wide operational network, at the Laboratories (UUCP: Unix to Unix researchers’ service. Copy Protocol).

Our narrative begins in 1980. What What’s happening in Europe? facilities did universities and Research into computer network researchers have at their disposal then to technology remains active. It splits into communicate between themselves or to two schools of thought: that of the accede, at a distance, to computer telecommunications operators, who

To conceive, construct and use complex instruments (here the ATLAS detector), particle physics as practiced at CERN requires the participation of hundreds of researchers, engineers and computer scientists: this culture of collaboration promoted the development of the Internet.

develop standards based on telephone impetus of Ira Fuchs who is in charge of network principles – this approach the computer centre of the City culminates in the X.25 standard, largely University of New York, they decide, due to the work of the Centre National on their own initiative, to interconnect d’Etudes des Telecommunications themselves and to offer an electronic (CNET) – and that of computer mail and news broadcasting service to scientists, illustrated in France by the their users. The name of the network is Institut National de Recherche en evocative: BITNet, Because It’s Time Informatique et Automatique (INRIA). Network. Its method of financing is The latter, led by , had very simple: each new centre starts to worked on solutions similar to those of seek a centre that is not too far away their American colleagues Bob Khan and already connected, and connects and (see article by Louis itself at its own expense. BITNet Pouzin p.32 in this magazine). If becomes operational at the beginning of experimental networks have been put in 1981. Most of these centres use IBM place with success in Europe, such as computers, and it is without surprise the French INRIA Cyclades network, that one sees IBM actively supporting no ambitious plan exists for the this initiative. In parallel, American deployment of either national or multi- university computer science research disciplinary infrastructures, i.e., open to departments interconnect themselves as all universities and researchers. Except well, by setting up the CSNet for two countries: in and in (Computer Science Network), under the Great Britain, work has begun on the initiative of Professor Larry Landweber one hand on the development of the of the University of Wisconsin. In required technology – in fact, everyone Europe, the British initiative, JANET – is inspired by the Arpanet technology, which starts to offer its first services – but nobody thinks to use it directly - is echoed. Germany creates a structure and on the other hand its baptized DFN (Deutsches implementation in a national network ForschungsNetz), SUENet and FUNet (UNINETT in Norway and JANET in greet the day in and in , Great Britain). During the period 1980- soon to be followed by SURFNet in the 1983, the situation becomes more Netherlands. They are, above all, complicated in the United States, and, development structures, not yet in Europe, awareness slowly dawns. operational networks. On the other side of the Atlantic, university computer centres, most of The year 1983 is a turning point for which are not connected to the Arpanet what will become the world-wide network, can no longer wait. Under the Internet. Whilst most European

2 countries do not yet have a network, the offers to the end user a tangible service first efforts at world-wide coordination which is called an “application”, e.g. start. Curiously, the very first meeting electronic mail and is implemented by ever to be held to envisage a general “application protocols”. As of 1983, world-wide research network is not held the Arpanet network, the ancestor of the in the United States. It is held in Oslo, American Internet, will use protocols in July 1983 (see box “First Steps”), at called TCP/IP for their transport the initiative of Larry Landweber, that technology. Their success was such exceptional visionary. Oslo inaugurates that, for a number of years, their name a series of events that Larry will will replace that of the network. One organize – the 1984 edition is held in talks of the “TCP/IP Network”. Paris – and which will be transformed, in 1991, into official In fact, TCP/IP contains two distinct In July 1983, the conferences of the Internet protocols: The IP (Internet Protocol), first meeting Society (ISOC). Once again, on the one hand, which specifies that where the future the impetus came from the data flow sent out by one computer world-wide universities and public sector to another must be split beforehand by research researchers who, without any the source computer into individual network is explicit mandate from their blocks, known as packets, and on the elaborated is institute, meet to lay down other hand, TCP (Transport Control held in Oslo together the base of a global Protocol) which allows the two initiative. The Oslo meeting communicating computers to detect if allowed not only to verify a common any packets are lost and if desire to create a trans-national retransmission is required. The network, and not just national islets, but specificity of IP is essential: packets also brought to light one of the major are independent of each other, as a letter difficulties that this enterprise would that is posted is independent, even if it encounter: the deep divide concerning is part of a more global correspondence. the choice of technologies. Now the Since packets are independent - just like most painful phase begins: the protocol letters sent via the postal service – they war. must each carry the address of the remote computer. Inside the network, The cannot be packets are forwarded by relays known dissociated from that of its technology. as routers, connected one to the other, Communication between computers just like the mesh of Post Offices and requires the use common rules and a Sorting Offices. It can thus happen that common language. These rules are an Internet packet, following a different called communication protocols. They route, arrives at its destination before broadly divide into two types. The first another that preceded it at the time of type is designed to ensure the transport emission. It is this principle, known as of raw information data between two the principle of connectionless network points, independent of their communication, that gave rise to the significance or of their use: These rules liveliest confrontations between a are implemented by “transport fraction of Europe and the United States protocols”. The second type of rule

3 During this epoch there are three types open to all: the IETF (Internet of Internet application Engineering Task Force. The protocols and this will Any individual or Internet designers set up this remain so until 1991: group of individuals, structure so as to benefit from SMTP for electronic mail, from a mere student contributions from all over the FTP for the exchange of to a group of experts, world, from universities, files and Telnet for the can propose an idea. researchers and engineers in connection from a terminal the public or private sector. to a computer at long distance. Thus, There is no need to belong to a no Web! These Internet protocols have Standards Organisation in order to two properties: they are open, i.e. they propose an idea, an improvement or a do not belong to any particular new norm: any individual or group of manufacturer and can thus operate on individuals, at whatever rank or any type of computer; they are public, qualification, from a mere student to a i.e. everyone has access to their group of experts, can propose an idea: specifications. In fact they are it will be analysed by volunteers from delivered via a light, informal structure, the IETF and will be retained if it’s considered pertinent. This work method, focused on the individual and, by extension, based upon a certain notion of individualism – the author’s name remains attached to his/her contribution – is fundamentally different from the hierarchical procedures in force in official Standards Organisations such as ISO (International Standard Organisation) or ITU (International Telecommunications Union) which, historically, promulgate the rules for telephony, television and radio communication.

While the American Internet refines its protocols, at the beginning of 1980 ISO and then UIT elaborate a set of rules baptized OSI protocols (Open Systems Interconnection). The objective is to promulgate rules that are open, i.e. compatible with any type of computer, and covering the whole spectrum of At the time when this drawing was sketched (1988) some people protocols, from transport to foresaw a division between world technologies: Internet in the United application protocols. Finally, States, OSI in Europe. In this model, the two sides would have the same objective as that of the communicated via gateways. Internet! The technological

4 divergence is at its greatest at the level of transport protocols. Influenced by Very rapidly, a significant fraction of telecommunication operators, who see the budding European networks chooses the future computer-based OSI technology as their communication services as an extension strategic direction. This If European industry of the telephone service, ISO and UIT is notably the case in adopts the OSI opt, in priority, for a connection- Germany and in Great approach, without oriented technology, as opposed to the Britain. As products do any soul-searching, Internet’s connectionless approach: not exist, apart from the American industry is thus, before an information packet can X.25 transport system, more circumspect be sent, the two computers must have temporary protocols are established a connection beforehand, developed: Great Britain, for example, i.e. the equivalent of a telephone call. designs and develops a set of software, If necessary, the invoicing will be based called Colored Books, on the JANET of the duration of this call. The network. This strategy in favour of the proposed standard, known as X.25, has OSI standards is encouraged both by the already been in existence for ten years,. European Commission and by a At the applications level, ISO and ITU majority of European Governments. also promulgate standards that are They perceive here a technological different from those of the Internet: domain where European industry is X.400 for electronic mail, FTAM for deemed to be significantly ahead. In the transfer of files. A difficulty arises fact, if the computing and due to the fact that at this epoch no telecommunications industry adopts the implementation of these protocols exists OSI approach without too much soul- in the form of operational software. searching, even sometimes with Some, such as FTAM, will never see enthusiasm, the reaction of American the day in the shape of a product! industry is more tempered. They

First steps The very first meeting about the idea of creating a universal network was held in Oslo at the beginning of July 1983. The initiative was due to Larry Landweber of the University of Wisconsin who directed the American network CSNet. The local host was Professor Rolf Nordhagen of the University of Oslo, a wise and resolute giant. In total a dozen representatives from varying horizons participated, such as the American Ira Fuchs, Chairman of the BITNet network, and the Englishman Peter Kirstein of University College London. I represented CERN, the European laboratory for Particle Physics, a great consumer of computer science and telecommunications resources. The idyllic frame of the hotel Holmenkollen (photo) overlooking the fjord of Oslo will see a common resolution emerge but technological divergences will also be brought to light, some Europeans insisting that any solution should use the so-called OSI technology from the International Standards Organisation (ISO). Sixteen years later, the participants’ most marked souvenir is the garden of Rolf Nordhagen where discussions carried long into the evening, late, very late, in the Norwegian summer, waiting for a night that did not come.

François Fluckiger

5 support it (like the computer responsible for the setting up of national manufacturer Digital Equipment) but research networks. On one side, the often with reserve. What worries the EARN network often precedes them in Americans is first of all the complexity their own country, in particular in those of the OSI standards: they abound in which, having opted for a complete OSI options, to such an extent that so-called strategy, must wait for product functional standards are invented, availability. On the other side, EARN specifying which part of the standard has appropriated the only name in must be used and under what English that perfectly describes what circumstances. The development costs national networks would have liked to do not leave them indifferent. On top of establish together: a European this the delay between two consecutive Academic and Research Network. But versions of the standards, from two to the major reservation concerns the four years, is thought to be ill suited to technology and its industrial the dynamics of their industry. implications: on the one hand, EARN uses proprietary protocols and, on the At the end of 1983 reaction to an other, the firm that supports these impromptu event of crucial importance protocols is not European. It will take will crystallize the gathering together of 18 months before reaction sets in. effort in Europe towards promoting OSI technology. Following the undisputed 1984 and 1985 see both the success of the BITNet network in the strengthening of the EARN network as United States, IBM proposes the a true trans-national communication creation of a similar network, based like tool and the multiplication of national BITNet on IBM’s own communication research network initiatives, apart from protocols, to the major European a notable exception, France. With the academic computer centres. IBM support of the European Commission, proposes to finance the connections the first informal conference of national between centres for a period of three research network representatives is held years. Herb Budd, on IBM’s side, is the on 14 May 1985. It is coordinated by initiator and overseer of this project. A James Hutton and Paul Bryant both of foundation meeting is held at CERN whom are British and is chaired by (European Laboratory for Particle Peter Linnington, the Director of Physics) in Geneva, during which the JANET. The outcome is a two-stage Englishman, David Lord, responsible at plan. In a first phase the objective is to this time for the CERN network, is harmonize the technologies that are elected Chairman of the Management used by various national networks and, Committee of the new Organisation: in a second phase, to interconnect them. EARN (European Academic and The problem is that in a number of Research Network). countries the national network’s existence is not yet tangible, apart from The network starts operating in the an administrative structure: to be beginning of 1984 and covers most truthful, often only the EARN national European countries. For the first time, branch exists! Above all, the universities and researchers have a technological orientation is clear: trans-national communication tool. national networks are converging Although limited to electronic mail and towards the use of OSI standards, and news broadcasting, EARN’s success is their interconnection will also use these immediate. However, the initiative is ill standards. A decision is taken to create perceived by a lot of organisations a common organisation that will

6 assemble national networks to pursue needs of computer scientists in this goal. The most appropriate European public research centres who acronym in English is no longer use the UNIX operating system and available and a French acronym is who need to use a trans-national chosen: the new organisation will be network. Moreover, at this epoch called RARE (Réseaux Associés pour la EARN is centred on IBM and DEC Recherche Européenne). It will take a systems and RARE does not yet have an year to legally set up the organisation, international network. The main EUNet and RARE will see the day on 13 June partners are the Royal Institute of 1986. It’s a Dutch association based in Technology (KTH), Stockholm, and the Amsterdam. Its initial financing is National Research Institute for ensured by the Dutch Government, and Mathematics and Computer Science then by the European Commission’s (CWI) in the Netherlands, where Daniel Directorate General XIII. Peter Karrenberg works and who will become Linnington is its first Chairman. the network coordinator, and INRIA which in France will play a major role in the national network. The network The period 1986-1988 is still marked supplies message and news services and by the confrontation between is connected to its American parent. In technological strategies. Indeed two 1988 EUNet announces its conversion distinct fronts will be active. On the to Internet technology. first, EARN’s IBM protocols confront OSI protocols whose products are A second front in the technology starting to become available (especially conflict has been open in Europe since X.400 for electronic mail). In a gesture 1986. This time the OSI and Internet of conciliation the Irishman Dennis protocols are in opposition. That year, Jennings who has replaced David Lord RARE proposes to the European at the head of EARN, proposes to Commission that an ambitious project Arriving from replace the EARN transport to establish an OSI network infrastructure by an OSI infrastructure for European research be nowhere, a technology infrastructure. financed. The project is known as research network This solution is needed to COSINE (Cooperation for OSI appears on the obtain the support of a new Networking in Europe). The task is international sponsor, Digital Equipment, immense as nothing exists, neither scene and who takes over from IBM. products nor certain protocols. A long converts to In fact, the international specification phase begins and is Internet in 1988 EARN network will remain finalized in the autumn of 1988. The the major method of project endorses a new Yalta. It international communication between implicitly recognises that the USA will European universities and researchers no doubt use Internet protocols, while until the end of the decade. Europe intends to use OSI protocols. Two worlds would exist, side-by-side, In the meantime, another network, with different network languages, and arriving from nowhere, appears on the obliged to communicate via gateways - international scene, without being a kind of automatic translation system. invited by any official body. Its role Ever since the use of different norms will be decisive. This is known as for, amongst others, television, EUNet (European UNIX Network) and technological schisms are considered a is a replica of the American USENet normal part of life. The implementation created in 1979. Its birth is due to the phase of COSINE begins in 1989 and

7 will be consolidated by the setting up of technology. It becomes difficult to an international transport network, IXI, refuse this demand as the criticisms based on X.25 technology. It will never aimed at EARN concerning the use of a be finalized, stopped in its tracks by the “proprietary” technology cannot be surging wave of Internet technology in levied against the Internet, an open Europe! technology if ever there was one. Only the industrial argument remains: In effect, Internet technologies (TCP/IP Europe is perhaps better armed in OSI together with file transfer and Internet technology than the United States. messaging) have, in the meantime, Events will show that this is not true. made a violent entry onto university More and more it appears that the heavy campuses. The technology is simple, OSI technology favours major firms efficient, is integrated into UNIX-type able to invest in battalions of engineers operating systems and costs nothing for so as to be ready to offer a product if the users’ computers. The first necessary. However in this domain this companies that commercialise routers, is not a characteristic of European such as Cisco, seem healthy and supply industry which on the contrary seems good products. Above all, the more likely to develop lighter Internet technology used for local campus products that require lower investment networks and research centres can also for the creation of expertise. As a matter be used to interconnect remote centers of fact, the firms that win the Internet in a simple way. The fundamental market, like Cisco, are small. Simply, difference with the previous systems they possess the Internet culture, are lies in the fact that from now on, users interested in it and, notably, participate exchange messages or transfer files in IETF. without having to worry about the different makes and types of computer. Towards the end of 1980, changes in Certain national networks do not opinion will be felt in the national hesitate and adopt Internet technology networks that are the most reticent very early on, not only for their towards Internet. In Great Britain domestic communications but also for JANET starts to convert. It is time for their interconnections with other RARE, that made OSI technology its European countries or the United States. strategic guide, to react. Three experts, In this way, the Scandinavian countries the Finn Lars Backstrom, the set up infrastructures based entirely on Englishman Brian Carpenter and the Internet, forming a federation thanks to Frenchman Guy Pujolle receive a the NORDUNet organization, and mandate to audit the situation. The interconnecting themselves to form the three sages hand in their paper on 22 first regional Internet network in January 1990. They recommend that Europe. On top of this, they connect RARE accepts the TCP/IP technology. directly to NSFNet (National Science At its conference in Vienna on the 1st Foundation Network), the central of February RARE adopt a resolution Internet network in the United States. that has remained famous. It is an The Dutch network SURFnet is also historic revision, painful for some: one of the first in Europe to convert to "RARE, without putting into question its Internet protocols. OSI policy, recognizes the TCP/IP family of protocols as an open multi- The European networks that continue to vendor suite, well adapted to scientific use the OSI strategy are now confronted and technical applications." by a strong internal demand for Internet

8 The political path has been freed, address, a unique identifying number. everything can now happen very Historically, the NIC (Network quickly. Information Centre), an American organization, allocated addresses for the Before reaching this point, Europeans whole world. The Americans, who believed in the Internet transport sometimes accused of wanting to technology had to maintain credible dominate the Internet world, on this contact with the rest of the Internet occasion showed great respect toward world still under construction. the Europeans. A few months later a NORDnet and EUNet, amongst others, non-profit-making European tackle this task. A world-wide organization saw the day – it’s mission coordination structure had also seen the was to grant Internet addresses to day in 1988: CCIRN (Coordinating European computers. Its creator, Daniel Committee for Intercontinental Karrenberg, gave it a French name: Research Networks). This was the Réseau IP Européen, whose acronym source of an important advance in (RIPE) is a play on words in English… Internet autonomy. In fact, the Americans, in the memorable meeting As of 1990 the European Internet of this committee (see box “The transport infrastructure grew rapidly. Abandon of Power”), insist that Europe An important factor in its acceleration, takes over the attribution of Internet six years after EARN, was a second addresses for computers that are IBM initiative that, again under the connected. It is a well-known fact that impetus of Herb Budd, proposed, for a each connected computer must have an limited period of three years, to finance a set of international connections and Abandon of Power switching equipment in order to create a major European network based-on The third meeting of the Intercontinental Committee for the Internet technology, with a higher Coordination of Research Networks (CCIRN) took place in a transfer rate than that already in summer resort of Western Virginia, sad and grey in this existence and to connect it at high speed month of October 1988. The Americans turned up in force. Bill Bostwick, of the Department of Energy was the Chairman. to the American NSFNet network. The Barry Leiner of the Department of Defense was present, network was organized in the form of a together with Dave Clark of MIT, one of the great Internet great European star around one point of architects. The European representatives were thin on the connection to the United States, and ground: a German representative, a British representative CERN was chosen to play this hub role. and myself. Towards the end of the meeting the Americans In February 1990 the first transatlantic became insistent: “It is essential that you, the Europeans, set Internet link with a transfer rate of l.5 up as far as possible a structure to allocate Internet addresses in Europe. It is not good that we keep doing it for you.” My million bits per second, entered into two colleagues did not seem too concerned, replying that service between CERN and Cornell they did not foresee the use of Internet rules in Europe. I University on the east coast of the thus took it upon myself to initiate this operation. On my United States. return, I sent an electronic message to about one hundred European engineers and computer scientists whom I knew to At the beginning of the decade, the be interested in the Internet. I asked them, if they wanted to European Internet is essentially a contribute to the coordination of addresses, to contact Daniel network used by universities and Karrenberg at EUNet in Amsterdam, who had undertaken to coordinate everything. The number of replies outstripped researchers. Some countries have our hopes! already adopted its technology: the Scandinavian countries, the François Fluckiger Netherlands, Spain, Ireland, Switzerland, Austria; some are in a

9 transition phase (Great Britain). above all, the Internet connectionless International communications are made principle destroys the practice of either via entirely Internet networks invoicing individual calls. Interested (EUNet, EASInet), or via the IXI private users turn towards research network which uses X.25 technology networks. Then begins a transition (soon to be replaced by the Internet period of three to four years where technology network EMPB). some national or international networks agree to connect non-university In France, the situation is more academic users on a temporary basis. confused. As yet there is no single Some will even cross the Rubicon and national network, but a stack of varying will become the first Internet Service types of network, generally using Providers (ISPs) in Europe. This is the Internet technology (the French section case for EUNet, renamed EurOpen - of of EUNet, the French part of EARN, a which Glenn Kovack becomes the few disciplinary networks such as that executive director- and is also the case of particle physicists). On the for some of EUNet’s national branches. international side, as of 1988 INRIA at In parallel new companies are created Sophia-Antipolis had put into place an here and there to offer Internet Internet connection with the United interconnection services, offering States, at the initiative of Christian general local or regional coverage. In Huitema. NSF, one of the project’s Great Britain PIPEX is one of the first. partners, will reinforce its contribution The clients of these first access to the liaison, which will allow INRIA providers are not satisfied with just a not only to act as a doorway for a part local service, they want world-wide of France to access the United States, access, both commercial and academic but also as an hub point for a number of – above all academic, as at this time, the Internet connections with great majority of information is stored Mediterranean countries. It is only at on university and research sites, and the end of 1991, notably under the there are practically no commercial impetus of Christian Michau of the servers. However, there is no European CNRS, that Renater, the national international network in existence network for education and research, will whose statute allows it to interconnect enter into service, catching up on lost isolated commercial access providers to time by the immediate setting up of a give them access to the academic part of powerful infrastructure. Internet. As for the private sector, it seems unable to set up such a network If the Internet is essentially an academic on its own. network at the beginning of the 1990s, the non-academic world is beginning at The public part of the Internet, i.e. this epoch to take an interest, its main universities and research centres, will objective being electronic mail. As for the last time come to the aid of the there is no commercial Internet private sector to set up the appropriate network, European telecommunication infrastructure: Ebone (European operators continue to disdain Internet Backbone). In 1991 a certain number of technology: which has the reputation of research network managers, such as the being unreliable and IETF meetings, Dane Frode Griesen, Kees Neggers, where engineers in sandals and director of the Dutch network computer scientists with long hair (such SURFNet, and the author of this article as the late ) are side-by-side, realise that the task must be achieved by hardly inspire confidence; last and establishing an international network of

10 mixed interconnection, i.e. open to However, the history of this period is everyone, publicly financed academic not finished. Whilst Internet networks or private commercial technology was developed by networks. The principle is simple: the universities, politics will now enter into network will consist of international action. During a conference on links and switching nodes; each telecommunications in Buenos Aires in participant will connect to the nearest 1994, Vice President declares: node; the overall cost will be divided “I have come here, 8000 kilometers amongst all the participants; no profit from my home, to ask you to help create will be made. This set-up is governed a Global Information Infrastructure”. by a simple Memorandum of Every country starts to frenetically Understanding (MOU, the first draft of produce plans for a national information which will be written by the author of infrastructure. But what exactly does it this article). all mean, what are these information highways that everyone is talking about, The network starts up at the end of 1991 in the G7 or in television studios? Is it with 14 member organizations. In the the Internet? It doesn’t seem so. autumn of 1992 ten other members join Reports talk about optical fibres in the Ebone consortium, and Frode private homes, about servers providing Greisen becomes its Director General. television programmes on request, The circle is thus closed and private about ATM transport technology, about Internet networks can fly on their own. bi-directional satellite antennae. But Internet Service Providers will not about Internet, except perhaps to progressively develop, later on mention that it is not the right solution, telecommunication operators will enter but is may be at best an intermediary onto the stage, and interconnection step. And who is supposed to create and points for access providers will bloom finance this new infrastructure that will almost everywhere in Europe, cover the world, in not Internet? The commercial users will quit public private sector. Five years later, one is networks which will completely rethink forced to admit that the information their mandate – to serve universities and highways have disappeared: only researchers – and the British statutory Internet, which overtook as the unique entity DANTE will be created which world network, remains. will be mandated to supply the European transport network. This narrative cannot be completed without evoking one of the essential Internet technology having come out on contributions to Internet technology top, the conflict between technologies, which came from public research: the will calm down (apart from a few fits World Wide Web. Is it necessary to and hiccups) the most active adversaries remind you that this Internet application of TCP/IP will discover a new passion was invented by Europeans, in the heart and will reinvent their past as veteran of Europe, in its biggest scientific defenders of Internet which will laboratory, CERN? In March 1989, sometimes bring them honours and Tim Berners-Lee proposed an promotions. RARE and EARN, rival information management project to organizations in the 1980s, will even CERN which did not receive much merge in 1994, and TERENA is born attention (see the interview with Tim from this union. Berners-Lee in this magazine). He had to re-submit his project a year later and only then did he succeed in obtaining a

11 few resources in order to start up the presentations were made. As the development of the system. A software was freely made available by prototype of the web-browser was CERN, then at the heart of the demonstrated at CERN at the end of European Internet, its growth was like a 1990, and, in 1992, the first public flash of lightning: the number of web Euro Internet in 1991: A gigantic star around CERN At that time, 85% of the international bandwidth installed for the Internet in Europe terminates in the Computer Centre building 513 of CERN.

12 servers grew from 26 at the end of 1992 transport network developed at the same to 200 in October 1993. In the summer time, and that CERN, having the of 1994, Tim left CERN to create the greatest capacity for Internet transfer in W3 Consortium at the Massachusetts Europe, was at the heart of the network. Institute of Technology (MIT), keeping It was also luck -or good design- that a European branch, initially at CERN supplied the Internet with mechanisms and then at INRIA. Mark Andreesen unique in the history of had just created Netscape together with telecommunication. One of these Jim Clark, whilst in Europe the Web mechanisms, essential to the was of no interest to any functioning of the Web, is the fact that industrialist….. any connected system, any connected object can have a unique name, This way of using the Internet is today independent of its real address; as well merged with the network itself. an automatic system (Domain Name Shortcuts became necessary, the Web System, DNS) allows the and the Net are become synonymous. transformation of these names into This distortion of the initial reality addresses. It was also chance that which no longer differentiates between several cutting-edge technical skills the transport of information from what were gathered together with Tim is transported irritates the specialists. Berners-Lee, in particular a great On the bottom line they are right – knowledge of telecommunications and although…… The Internet has possibly of document structure techniques. In simply become the Web – as practically our more and more complex any communication between man and technological world, the machine can be made via a Web compartmentalization of individuals window – and the rest is just mere into specialities is a brake on creativity detail. and often progress is made by those who master more than one domain. The There are numerous reasons for the double invention by the same person of success of the World Wide Web project, the Hypertext Markup Language whereas many other technological (HTML) and of the Hypertext Transport projects have foundered. Like a lot of Protocol (HTTP) is an illustration that births, the coming of the Web was will mark the history of our society. perhaps a mix of chance and necessity. Too many projects seek their goal The method should not be forgotten. retrospectively. The Web fulfilled a Rather than just specifying the genuine need, that of a well-defined technology, which is the traditional community – the community of particle method used by standardization bodies, physicists. This community is made up Tim invented it and offered it to all. of ten thousand researchers and Rather than saying “this is how HTTP technicians, spread all over the whole works, it’s up to you to develop it and world. The planetary dimension was to test it”, he said, “here’s the software, thus present right at the beginning. It download it, it works”. And it worked! was just chance that the Internet

François Fluckiger January 2000

13