Internet Host: nic.cerf.net Directory: farnet Subdirectory: farnet_docs Filename: euronetconf-report Last Updated: Aug 5, 1991 REPORT ON THE 2ND JOINT EUROPEAN NETWORKING CONFERENCE Blois, France May 13 -16, 1991 by Richard Mandelbaum This report consists of 3 parts: I. OBSERVATIONS AND OVERVIEW II. THE CONFERENCE TALKS III. CONCLUSIONS In addition there is an Appendix containing some background material on RARE, which was adapted from their annual report. I. OBSERVATIONS AND OVERVIEW This is the second Joint European Networking Conference that I have represented FARNET at and it is still hard for me to give a good explanation of the European Networking Scene. The organizational, political and technical situations is much more complicated than in the United States. First of all, there are two organizations, each of which can fairly claim to represent research and educational networking in Europe. One is RARE (Reseaux Associes pour la Recherche Europeene),which is the association of European networking organizations and their users. The second organization is EARN, (European Academic Research Network) which is the European counterpart of CREN. RARE seems to be viewed as the official representative of Europe by the FNC. With the increasing regionalization of EARN; however, my view is that EARN is probably more representative of the end user networking institutions in Europe. In American terms, RARE is a mixture of some elements of FARNET, the NTTF and the IETF with a bit of Merit (in its role of managing the NSFNet) thrown in. Of course, the situation is made even more complicated by the fact that the members of RARE represent nations rather than states or regional organization as in the U.S. RARE places much more emphasis on technical/user services standardization issues than FARNET has to date and the annual European Networking Workshop seems to be a mix of IETF meetings, the annual NTTF National Net Conferences, and some of the FARNET meetings. The two major technical/political problems facing RARE, or more accurately, European networking in general, are: The intense controversy between the X.25/ISO CONS camp and the TCP-IP/ISO CLNP camp. There is nothing approaching general agreement on how networking should be implemented in Europe. For the first few years, after its founding in 1985, RARE was actively and vociferously in the X.25/CONS camp. TCP/IP was viewed as a dangerous heresy which had to be stamped out. At the Sixth International Networking Workshop held in Princeton in 1987, I remember hearing various prominent RARE types saying things such as "We will never permit TCP/IP in Europe." As more and more UNIX workstations with working TCP/IP implementations were sold in Europe, reality began to intrude upon theory. TCP/IP was alive and well and in addition it worked. More and more European Academic and Research institutions implemented TCP/IP and did internetworking between themselves using it. In addition, national networking bodies in Europe, especially NORDUNET became TCP/IP advocates. By 1990, RARE did a volte-face, and somewhat reluctantly adopted RIPE, the main TCP/IP body in Europe, as a RARE subgroup. Despite RIPE's becoming a RARE body, the issue is far from settled. Last year's meeting at Killarney was highlighted by a rather high spirited and vocal interchange between NORDUNET representatives and representatives of DFN, the West German National Network on the issue of X.25 vs. TCP/IP. In the past year, RARE began operating IXI, (International X.25 Infrastructure), a pilot X.25 backbone network. Traffic on IXI rose to about 5 gigabytes by the end of 1990 (about 1% of NSFNet traffic). Although IXI is a start at some sort of intra- European backbone it is not universally admired. The service, according to many of the delegates I spoke to is slow and not particularly reliable. It doesn't solve the need for TCP/IP LAN interconnection and with 64 kb pipes doesn't give anything which can be, in 1991, called high speed. That leads us to the second major European problem. The lack of a true high speed intra-European backbone. In order to understand the need for such a backbone, RARE organized the EEPG (European Engineering and Planning Group) at the Killarney Joint Meeting. The EEPG was charged to: (1) make a quick and coarse analysis of the present international European traffic flows and estimate its potential development; (2) make a quick survey of the existing international lines and resources used; (3) after a discussion of technical possibilities, choose available technologies on which a backbone could be based; for each of these have a group write a description as to how a backbone would look like using this technology; and (4) find out organizational possibilities on how a European research backbone could be operated. The EEPG Final Report was made public on May 2, 1991, and discussed at the Conference. The basic conclusions of the report were: Technical aspects. A choice should be made between an X.25 plus embedded protocols and a multi-protocol TDM backbone data communications network. Planning should then be started for the immediate implementation of a 2 Mbit/s pan-European infrastructure. In parallel an immediate start should be made to plan for the implementation of an ATM pilot project which will address the areas of future technological requirements and higher speeds. Organizational aspects. A study should be undertaken to determine the appropriate legal and financial organizational structure required to allow the formation (in the near to short term) of an independent operational unit which can establish the backbone service and manage the ATM pilot project, other future projects and user services. Unfortunately, those conclusions beg the issue and the lack of any procedure to reach a decision was not very well received. My own view is that slowly, but surely advocates of "multi- protocol solutions" are gaining the upper hand in the fight and we will see the beginnings of a multi-protocol European backbone in the not too distant future. Again, in U.S. terms of reference, I would say that on the network/transport side, Networking in Europe as about where the U.S. was in early 1988 with the exception that no contract for an analogue to NSFNet is in place. When that changes, there will be an immediate dramatic increase in intra-European networking and it will probably take the Europeans about 12 - 18 months to traverse the distance the NSFNet backbone covered between 1988 to 1991. With the lack of any organizational with the funding and clout of NSF, I don't yet see a strategy which will take the Europeans to a T3 backbone. The major difference between the U.S. and the Europeans approach to networking is: 1) the role of the EC in Europe and 2) the NSF's proactive approach in the U.S. The EC is completely focused on ISO. As a consequence of the Treaty of Rome, establishing the EC, International Standards must be the basis of any telecommunications effort which the EC supports. Unfortunately, until now, the EC has defined International Standards to be exclusively ISO standards. Furthermore the EC approach to networking is a typically bureaucratic one. Extensive studies are carried out of current and projected capacity increases, user needs are surveyed, funding strategies are assessed and then implementation plans are put into place. In the U.S., lines and services are upgraded because NSF feels that such upgrades will in the medium term help R & D and tech transfer. Ken King once remarked that trying to assess future bandwidth needs is like trying to decide whether to build the George Washington bridge based on a survey of the number of people swimming across the Hudson. In Europe, they do count the swimmers. However the rise of regional associations, such as NORDUNET, and the increasing activities of EARN are having their impact. However, I do believe a lot is going to change after December of 1992 when the last barriers to competition in telecommunications are slated to fall. The area where I think the European Networking Community is, at least abreast of us and probably ahead, is in the area of services on our network. There were a number of excellent presentations a the conference on X.400 and X.500 projects, which seem to be getting much more attention from national net managers in Europe than similar projects in the U.S. are getting from FARNET members. Similarly, the Cosine Information Services Projects are more advanced than corresponding efforts in the U.S. II. CONCLUSIONS The Conference itself was mediocre. A major difference between the Killarney Conference and this one was the lack of an EARN presence and the poor division of talks between plenary and workshop sessions. Nevertheless, it is important for FARNET to send representatives to RARE and in my opinion become a liaison member the way CERN is. Networking is an international activity and there are quite a few areas where we can learn things from RARE and EARN. Furthermore increased US participation in European networking events strengthens the hands of those groups, such as NORDUNET and EARN, who are pushing for multiprotocol networks and less slavish obeisance to ISO. The idea of a conference, such as the RARE Conference, is a very good one. The Conference is unlike anything I have seen in the U.S. At its best, it is a mixture of technical discussion (IETF), political discussions (National Net) and network policy/management discussion (FARNET) meetings. A similar conference in the U.S. perhaps jointly sponsored by the NTTF, IETF and FARNET would go a long way towards mixing the key network players in a more intimate fashion and perhaps lead to a stronger U.S. organization which can truly say it represents high speed networking in the U.S. At the very least, such a conference might lead to a better understanding among the leaders of the various streams in the U.S. networking community of what each stream is doing now and needs to do in the future. III. THE CONFERENCE TALKS I will arrange to have the abstracts of the conference available through anonymous ftp at farnet.org. I append some specific comments about a few of the talks I found noteworthy. Complete versions of the talks will appear in the forthcoming Autumn issue of Computer Networks and ISDN systems. 1. THE NETWORKED CAMPUS Bernard Levrat, Universite de Geneve, Switzerland ABSTRACT: Started in 1985, the Universite of Geneva network interconnects with optical fibers or phone lines a set of Ethernet LANs covering 87 buildings. It was started from scratch, ignoring existing equipment, connections and applications. Access to Ethernet through TCP/IP was part of the requirements of all the subsequent call for tenders, including large mainframes and interactive machines. After 6 years, 400 terminals, 1500 IBM-compatible PCs, 110 Macintosh, 300 workstations give access to many services including access to all our servers and their applications, data archiving, software distribution, e-mail, NEWS and access to national super-computers through SWITCH. Distributed printing and sharing of some 250 printers, most of them laser, present special problems. In addition to TCP/IP, our network supports DECnet, NOVELL and AppleTalk. Problems of connection discipline will be mentioned. The presentation will emphasize continuity and enhancement of network services, leaving the discussion of network management to other speakers. COMMENTS: The most interesting fact was that Levrat does global updating and servicing of the network centrally using FTP Networking Software. This was a plenary talk, showing the lack of logic in the division of talks between plenary and group session. The talk was fairly typical of the show and tell sessions many of us in the U.S. sat through in 1987/88. 2. WORKSTATION TO INTERNET: PROBLEMS, SOLUTIONS AND CHALLENGES Lee G. Caldwell, Novell, Salt Lake City, Utah ABSTRACT: As national and international research networks evolve, they must, of necessity use international standards. Different communications protocols dominate the desk-top for five reasons. First, workstation processors and memory limitations make it impossible or impractical to support the entire standards bases protocol stack. Second, workstation vendors introduce new services which can only later be reduced to standards. Third, workstation based networks require simplified installation, maintenance and management. Fourth, workstation LANs offer relatively low connection costs. Finally, workstation LANs must also support connections to proprietary based host systems. The low cost, transparency, high performance and reliability of workstation LANs has made them popular with scholars and researchers. As national and international research and education networks plan to offer services to a broader community of scholars, links to workstation LANs become essential. The diversity in workstations and LANs requires use of several integration tools in concert to achieve reasonable levels of transparent, high performance service of internetworking protocols. Various academic and research departments will likely need different levels of service and performance. This paper present solutions for connecting workstations to research and education networks. Both university developed (public domain) and commercially available solutions are discussed to assist organizations in selecting an economically and technically feasible mix of tools and services. A popular integration solutions supports multiple protocols in the workstation. The first solution for IBM PCs and compatibles uses a "Packet Driver" technology. Today many public domain and commercially developed solutions support both TCP/IP, various PC LAN XNS based or AppleTalk protocols concurrently. In addition, a number of vendors developed intelligent Ethernet boards with the TCP/IP protocol stack in firmware to reduce memory requirements for the PC. Second generation tools conform more closely to the data link level of the ISO OSI protocols and include the 3Com NDIS specification and the Apple/Novell Open Data-Ling Interface (ODI) specification. The NDIS protocol loads and unloads protocols as needed while the ODI protocol concurrently supports up to 32 protocols. Both specifications support popular interfaces for standard TCP/IP based applications like NCSA Telnet and others. As campus network evolve beyond flat Ethernets connected to the internet router, on- campus routing and/or bridging becomes an important consideration. In general, bridges are protocol independent and pass all packet formats. They may not support all network hardware options, however. Routers only pass the protocols they are designed to support, but offer advantages in management and support for mixing more network hardware types. Often, LAN systems offer integrated routing services that only support a single protocol. Client based solutions based on concurrent support of multiple protocols cannot operate unless bridges or multiple protocol routers are used. Universities may use a single multiple protocol router for each building or they may use a separate router for each protocol (such as the public domain KA9Q software running on a separate PC). A third approach uses intelligent add-on routers (based on Ethernet boards) for popular network operating systems. Gateways also extend higher level internetwork services to workstations, but do not sacrifice scarce workstation memory. Standard LAN protocols connect the workstation to a gateway server that translates the requests into appropriate calls for internetwork services. These gateways may be installed internal to a PC based file server or may reside on a minicomputer. Typically these products offer support for terminal emulation, file transfer and X.400 and/or SMTP mail. These products offer some management advantages when dealing with large numbers of casual internet users. New technology solutions include multiple protocol support in servers, protocol encapsulation (or tunneling), support for concurrent file services (NCP< NFS, AFS and FTAM), remote procedure call (RPCs), and layer specific gateways. UNIX System V, Release 3 STREAMS and Release 4 Transport LIbrary Interface (TLI) provided interesting tools to encourage support for multiple protocols in servers. Other multiple transport tools such as improved TCP/IP Sockets are evolving from companies and industry organizations. A number of new architectures allow servers to support several file systems concurrently and transparently. RPC tools allow developers to establish interprocess communications without concern for underlying transport and network protocols. Finally, layer specific gateways allow the use of a variety of high level services regardless of the underlying transport or network layer protocols. Taken together, these tools extend the reach of national and international networks to local networks without sacrificing the performance and transparency of the workstation LANs. While it will be come years before this objective is totally achieved, a number of tools using current technology, if applied in concert, can make significant contributions to the reach of national and international research and education networks. This paper will discuss these and other issues with technical details and examples of successful implementation by a number of colleges and universities. COMMENTS: A good quality vendor presentation, emphasizing what Novell was doing. Major stress on Novell's efforts to insure OSI/TPI capability and on the road to guarantee Inter- Vendor compatibility through extensive interoperability testing. 3. ORGANIZATION STRUCTURES FOR THE PROVISION OF INTERNATIONAL DATA COMMUNICATION SERVICES IN THE RESEARCH COMMUNITY Klaus Ullmann, DFN Verein, Berlin, Germany ABSTRACT: Starting with a short description of the underlying technical tasks the paper analyses in its first part the funding background for the provision of these services. In the second part some general political guidelines are defined and discussed. This leads then to the third part where the organizational structures and implications are described. Finally, necessary steps to achieve these new structures are outlined. COMMENTS: Technical Needs (1) Operation of Shared Gateways (2) European Connectivity Structure (IXI) X.25 (3) X.400 Backbone (4) Enhanced RIPE All of the above need funding structure and organizational framework. [European Research Networks are roughly on par with U.S. during 1987/88. At Campus level, about the same as the U.S.] EC has no mandate to finance the European Infrastructure! a. Example: Shared Gateway Service: X.400 - SMTP Gateway Funding 100 KECU/year b. Principles * Minimal Consensus - must guarantee application connectivity * Voluntary - no one forced to use a specific structure * Subsidiary - do not duplicate what can be bought * Customer Funding - those who use it pay (at network level, network's are customer) * Non profit * Open - any network can use if they pay * Steering - Those who pay, govern c. Role of RARE * Rare is a consensus finding body (political role). * Service organization is a task-oriented body with financial risks involved. New structure is needed, inside RARE, outside RARE! * Key issue is governance, vs. services! 4. ACCEPTABLE USE POLICY J. S. Hutton, A. Jeffree, Joint Network Team, Didcot, UK ABSTRACT: Joint Network Team is a necessary part of managing networking services for the research community that the services are operated against a definition of the purpose for which they are provided. Acceptable Use Policy can be considered under the following headings: * definition of the primary user community; * technical use, which covers both categories of use such as data, video and/or voice, and application protocols to be supported; * site connection requirements; and * prohibited use and the prevention of abuse; The paper will discuss the recently developed JANET Acceptable Use Guidelines with emphasis on those aspects that are considered to be of interest in the international context. Background JANET has usually been considered as a network established exclusively for the UK academic community. Following developments in other countries; however, it was recognized that significant benefit would accrue to all concerned if the network was opened up to the industrial research community. The first activity to achieve this goal was the development of a clear set of Acceptable Use Guidelines. These were developed by one of us (AJ) while working as a member of the Joint Network Team funded by the UK Department of Trade and Industry. COMMENTS: Must clarify who the network is for. Move from "pure" academic network to one involving industry (at least at research level) as well! The primary focus is on Higher Education and Research. Who is involved? How? Scope of Acceptable Use (1) Users and their use. (2) Applications supported - technical specifications. (3) Applications permitted. (4) Connection requirements - management and technical. Issues (1) Commercial Use (Totally different environment, commercial doesn't seem commercial, much more primitive.) (2) Privacy & Security (3) Legal Constraints (4) International Alignment (5) Song & Dance (Multimedia) (6) Charging Conclusion Must document acceptable use! 5. REPORT ON THE RARE/CEC SYMPOSIUM ON HIGH SPEED NETWORKING FOR RESEARCH IN EUROPE. Paul Van Binst, Universite Libre de Bruxelles, Belgium ABSTRACT: On January 24, 1991 in Brussels, 130 experts assembled to review the present and future of high speed networking for research in Europe. Among the participants were a majority of research networking organizations and universities, as well as a large number of representatives of high level research institutions (ESA, CERN), industry, governmental organizations, Public Telecommunications Operators and the European Commission. Fifteen high level presentations addressed such topics as * backbone technologies, * existing advanced networking implementations (wide area supercomputer networks, metropolitan area networks), * special requirements of space applications, * high-level industrial applications (aerospace, car manufacturing), * national initiatives for high performance networking implementations. The conclusion of this highly successful symposium was that the majority of advanced high performance networking projects and initiatives in Europe take place at the national and regional levels, but that there is still a crucial lack of international coordination. COMMENTS: Problem: Tariff & Structure (Political and Organizational). Some progress at National level. Major problem on Trans- National level. Need to work with PTTs. Must understand and convey idea that Computer Technology and Communication Technology will be tightly coupled in the future! 6. THE NSF X.400 PILOT PROJECT Allan Cargille, Rob Hagens, Alf Hansen, Lawrence H. Landweber, University of Wisconsin, Madison ABSTRACT: At the University of Wisconsin-Madison, a project team is working on an X.400 Pilot Project with a 2 year grant from NSF. The goal of this project is to support adoption of X.400 in the Internet. One of the main activities is to operate an experimental (1984) X.400 service with full connectivity to the Internet Mail world. The project is participating in the R&D MHS Service coordinated by the RARE MHS Project in Europe. The Wisconsin team is offering users of X.400 systems in e Internet community some of the services they need for a smooth introduction of X.400 side-by-side with the Internet Mail service. The paper is a status report from the project, and will present the preliminary conclusions after 6 months in operation. The aspects covered are: * Registration of X.400 addresses * X.400/RFC 822 address mapping * Deployment of RFC 987 gateways * Experience with PP as a transition tool * International R&D MHS connectivity (WEP operation) * Internal service procedures * Connectivity to public X.400 service providers (ADMDs) * Routing strategies * Technical innovations * National cooperations * Future plans COMMENTS: This was a very good presentation of the NSF sponsored X.400 mail handling project at the University of Wisconsin, Madison. The projects is being done in coordination with the RARE MHS Project and runs the U.S. X.400 gateways in Norway, Switzerland, Spain, Finland and France. Currently, all R&E X.400 mail into and out of the U.S. goes through the Wisconsin gateway which switched about: 2500 X.400 to X.400 11000 X.400 to RFC 822 1500 RFC 822 to X.400 Messages in April, 1991 The main problems revealed or encountered by the project to date is the lack of an X.400 infrastructure in the U.S. and lack of incentives to migrate from RFC 822 to X.400. The interest in FAX gateways and E-mail to FAX conversion seems to be changing this. The most critical technical needs uncovered by the project so far has been the need for: (1) meaningful organizational names; (2) global coordination of R&D message handling services; and (3) coordination of address mapping rules. For more information, do an anonymous ftp to mhs- relay.cs.wisc.edu and retrieve "pub/gen/how-to-join.txt" 7.THE COSINE INFORMATION SERVICE Graham Knight, Level 7 Ltd., Bracknell, UK ABSTRACT: Relevance: The presentation will be of interest to existing and potential end users of information services, information providers and members of established and potential Special Interest Groups (SIGs). CONCISE Overview CONCISE (COsine Network's Central Information Service for Europe) aims to provide a pilot pan-European Information Service to the COSINE and the European Industrial Research Community based on an Open System Environment and accessible through OSI protocols. CONCISE Objectives Many countries already have national information services. The service provided by CONCISE will compliment these national services by providing information Europe-wide and by providing information about these other existing services. Thus it will form the central focal point for users to obtain information about pan-European networking, research projects, products and services, as well as about COSINE itself. An information server will be developed, which will hold a database of information accessible either interactively, by X.400, or by FTAM. The system architecture will be portable so that other information services and special interest groups will be able to adopt it to set up their own information servers elsewhere in Europe. Knowledge gained from the running of the pilot system will be used to study the feasibility of providing a self-supporting information service in the future, as well as recommending ways in which the service might evolve. CONCISE Status Currently the project is in its early stages. A close liaison with the members of the User Information and User Support Group has been established and a requirements specification for the system has been agreed. Planning for the technical development and integration work has started. CONCISE Plans The pilot service is planned to be available from September 1991 initially through X.400 access only. By January 1992 interactive access by X.29, VTP and dial-up will be introduced and by the end of Q1 1992 FTAM access will also be available. The pilot service will run until the end of December 1992. It is anticipated that the recommendations made at the end of the project will make it possible to set up a self-supporting service for 1993 and beyond. COMMENTS: The main purpose of the project is (1) provide a European Central Focal Point for Information; (2) build a prototype for National Information Services and Special Interest Group info sharing in Europe; and (3) to determine whether Information Services of this type can become self-supporting. A reasonably good talk with some far-reaching implications for what FARNET and its members should be looking at today. The COSINE scope extends significantly past transport services and into European National Networks looking seriously at playing a role in the provision of Information Services. 8.GLOBAL HIGH SPEED WAN ARCHITECTURE FOR THE 90S Remi Despres, RCE, Cergy Pontoise, France ABSTRACT: Several types of architectures are considered for high speed data networks of the 90's. They are likely to coexist. In this context, users and network operators should preferably understand the relative merits of these architectures. The purpose of this paper is to contribute to this understanding. In the 80's, the worldwide research community network, which expanded successfully, was based on a connectinless network service (CLNS), while the worldwide public data network, which also expanded successfully, was based on a connection oriented network service (CONS). For the 90's two additional architectures, frame relay and ATM, are envisaged. They are based on connection oriented packet mode, but, unlike with CONS, data sources are not subject to flow control from the network. High packet switching rates are thus achievable, but users can be subject to wide variations of queuing delays and/or to high rates of data losses in the network, when traffic conditions are not favorable, in particular if some other users are ill behaved. (ATM, however, does not suffer from these dangers when it is used only for multiplexing synchronous traffic, whatever the speeds). This paper explains why the CONS remains a particularly attractive basis for high speed shared data networks: (1) Exercising flow control on data sources is the only way to guarantee stable service quality under highly varying traffic patterns. (2) Low cost technology exist for 2Mbit/s X.25 networks. (3) High speed encapsulation of connectinless traffic can be performed at the WAN entry point. (4) For higher speeds (typically 34, 40, 140 155 Mbit/s), optimized packet formats and protocols, suitable for high speed operations on LANs, can readily be designed. (5) For bulk data transfers, CONS allows for very long packet lengths, with automatic fragmentation and reassembly of packets inside networks, which allows high data rate switching at low cost. (6) With high speed CONS, interworking between higher speed WANs and existing public WANs is straightforward. COMMENTS: A typical vendor type talk. Pointed out all the advantages of X.25/CONS and none of the disadvantages. Dismissed CLNS more or less out of hand. 9. FROM MEGABITS TO GIGABITS: A REVIEW OF POSSIBLE SCENARIOS Francois Fluckiger, CERN, Meyrin, Switzerland ABSTRACT: Over the last 3 years, certain segments of the European Academic and Research community have introduced international megabit lines to support international collaborative research programs. This allows new styles of working, especially in scientific fields where intensive numerical analysis is required and a large amount of experimental data is produced. This poses a dual challenge for the coming 3 years. The first is to actually trigger the use of modern and innovative network services to fully exploit the benefits of such a megabit infrastructure, and not simply to load it with an accumulation of rather conventional services. This includes not only pure data services but also multi-media facilities and group or inter-personal communications based on voice and video. The second is, in parallel, to pave the way for entering the gigabit era. This implies evaluating the multi- hundred megabit and gigabit transmission and switching technologies, planning their introduction, and ensuring the megabit technology used allows for a smooth transition. The presentation aims to intimately bind theory and practice, and will be based on the practical experience on mega/gigabit services and techniques gained within the European Particle Physics Community. The talk will review the megabit WAN infrastructure in use, its technological trends, and will discuss the current and expected usage; it will then survey the gigabit technologies and standards for communications and switching, will discuss their impact on resulting services, and present technical scenarios for progressive introduction in typical computing intensive research communities in Europe. COMMENTS: Among the best talks at the conference. Among the best presentations I have heard anywhere about planning and implementing a move to high bandwidths . CERN is the European Organization for Nuclear Research and is responsible for running the High Energy Physics facilities based at CERN headquarters in Geneva, Switzerland. (I believe their facilities are roughly on a par with the DOE sponsored Fermilab facilities in Batavia, Illinois.) CERN is the largest user of research bandwidth in Europe. In the EEPG report, CERN has about 12 Mbit/s in bandwidth accessible with the second largest being Switzerland with about 5 mbit/s. CERN has more bandwidth than France, Germany, Italy, England and the NeNeLUX countries combined! Fluckiger talked about CERN's plans to migrate to the new ISDN, Frame Relay and ATM technologies from the perspective of someone responsible for the running of production network services for a major scientific research lab. I think it would be very worthwhile for FARNET to arrange a talk by some members of the Fermi or Argonne HEP communities about their network plans. 10. CHEOPS: REALLY USING A SATELLITE J. Altaber, S. Cannon, B. Carpenter, D. Davids, C. Isnard, J. M. Jouanigot, N. Stheneur, CERN, Meyrin, Switzerland P. L. Forsstrom, CSC Helsinki, Finland M. Nordberg, M. Voipio, SEFT, Helsinki, Finland G. Barreir, LIP Lisbon, Portugal P. Veiga, INESC Lisbon, Portugal S. Dedoussis, C. Touramanis, University of Thessaloniki, Greece O. Koudelka, Technische Universitate Graz, Austria M. Hine, Geneve, Switzerland ABSTRACT: CHEOPS is a collaboration between CERN, LIP and INESC (Portugal), SEFT (Helsinki) and four Greek institutes represented by the University of Thessaloniki. Its objective is a sustained experiment in the use of ESA's OLYMPUS satellite to demonstrate that the unique properties of satellite transmission can be used to distribute massive quantities of scientific data for remote analysis. Previous experiments (STELLA, HELIOS) have shown that achieving high-speed point-to- point data transmission by satellite is a solved problem. CHEOPS aims to benefit from this experience by concentrating on three aspects: efficiency, access to geographically remote regions, and integration of the application of the satellite into the general computing environment. The four initial CHEOPS ground stations will have 3m antennae, 8 W radio-frequency amplifiers, and modems capable of bit rates up to 8 Mbit/s. They will access the Ku-band transponders of the "specialized services" payload of OLYMPUS, mainly on an overnight schedule when satellite time will be free for our experiment. At each site there will be a UNIX- based file server, integrated in a suitable way into the LAN and computing environment. During day shift, users (physicists) will make interactive requests to the CHEOPS system to transfer physics datasets, and the datasets will be staged onto the CHEOPS disc space. At night, links will be established from CERN to each remote site in turn, and the datasets will be spooled automatically from source to destination. As far as possible, the process will require no operator intervention. Initially standard TCP/IP protocols will be used, with multiple parallel transactions to improve efficiency. Later, optimized protocols will be used in an attempt to maximize throughput. A successful experiment on three fronts (geography, efficiency, and integration) would clear the way to possible use of commercial services, assuming economic tariffs and a liberal regulatory environment. COMMENTS: A good talk about a real 8 Mbit/s production service being piloted at CERN. Among the better talks I have heard about using wide area high speed networks in the scientific production fashion. APPENDIX BACKGROUND RARE, Reseaux Associes pour la Recherche Europeene, is the association of European networking organizations. The aim of RARE is "to foster cooperation between these organizations in order to develop a harmonized international data communications infrastructure." RARE has four types of members: (1) Full National Members The full national members are representatives of the following twenty European countries: Austria Iceland Spain Belgium Ireland Sweden Denmark Italy Switzerland Finland Luxembourg Turkey France Netherlands United Kingdom Germany Norway Yugoslavia Greece Portugal They are the voting members and formally constitute the association. (2) Associate National Members This group consists of representatives of the national research networks in other countries which support the objectives of RARE. Current members include: Czechoslovakia Hungary Poland Israel Republic of Korea (3) International Members These are international organizations within Europe which support the objectives of RARE and are closely associated with the use, coordination and provision of a European data a coordination infrastructure. Current members include: CERN European organization for nuclear research. EARN European Academic Research Network NORDUNET Union of Scandinavian Research Networks ECFA European Committee for Future Accelerators ECMWF European Center for Medium-Range Weather Forecasting ESONE European Standards on Nuclear Electronics EurOpen European Unix Users' Group (4) Liaison Members Consists of organizations which are involved in networking and related matters with whom RARE considers it important to have close and continuing contact. Current members include CREN. The primary activities of RARE are: I. TECHNICAL RARE sponsors seven working groups which are responsible for developing coordination and cooperation in technical areas. The subject areas of the Working Groups are as follows: Working Group 1: Message Handling Systems - create and promote a European infrastructure for a message handling service within the European research community, with connections to the global environment. Working Group 2: File Transfer, Access and Management - prepare for an efficient transition to FTAM - monitor the functions and features of the base standards and profiles - define and conduct interoperability tests. Working Group 3: Information Services and Directories - promote the establishment of information and user support services in thc RARE countries as well as directory services based on the X.500 standard. Working Group 4: Network Operations and X.25 - coordinate the exchange of management information between network service providers - address the related technical issues. Working Group 5: Full Screen Services - promote the use of OSI conformant Virtual Terminal Protocols - conduct pilot activities - track the related technical developments. Working Group 6: High Speed Communications and ISDN - address the technical and management problems related to high speed networking Working Group 8: Management of Network Application Services - address the problems of accounting, charging and costs distribution, and security and authentication. In addition, RARE sponsors: RIPE (Reseaux IP Europeans) This is a RARE body which is responsible for European TCP/IP coordination activities. RIPE's aim is to ensure the coordination necessary to allow the operation and expansion of a pan- European IP network. As such, RIPE acts as a forum for the exchange of technical information as well as for the promotion and coordination of interconnection of IP networks within Europe and to other continents. EEPG (European Engineering and Planning Group). The EEPG is charged with the study, preparation and planning of a European data communications backbone. Its final report was published about two weeks before the "Blois" conference. II. PROJECTS A. The Eureka Cosine Projects. The European COSINE (Cooperation for OSI Networking in Europe) consisting of Austria, Belgium, Denmark, Finland, France, Germany, Greece, Ireland, Italy, Luxembourg, Netherlands, Norway, Portugal, Spain, Sweden, Switzerland, the United Kingdom, Yugoslavia and CEC (Commission of European Communities) have funded RARE to manage the Eureka Cosine projects with the aim of establishing a pan-European OSI standard computer communications infrastructure for those working in academic, industrial and public sector research and development. The key projects being worked on are: Project P1.1 is the establishment and operation of a pilot FTAM gateway to North America. Project P2.1 is the investigation and establishment, in collaboration with national directory pilot projects, of a pan- European pilot X.500 Directory Service. Project P2.2 is the setting up a pan-European information service, facilitating the cooperation between national information services and offering useful information (for example, on products, national networks and contacts, and international projects) to the European networking community. Project P3 aims at establishing an OSI-based networking infrastructure and associated services to enable all European researchers to benefit from improved communications and access to information. Project P6 aims to arrange for different vendor products to be conformance tested by existing European test centers, and to subject those certified products to interoperability tests for a trial period with the intention of exposing any remaining incompatibilities. Separate sub-projects are identified for FTAM (P6.1), X.400 and X.500 products (P6.3). Project P.9.2 is to support the introduction of ISO-VT services and demonstrate the integration and complementary nature of ISO-VT and X-Windows. B. The MHS Pilot Project. In 1987 RARE undertook a pilot project aimed at the creation of a broadly based pilot infrastructure of emerging electronic mail implementations of the X.400 standard. The project resulted in more than 100,000 users connected through 530 computer installations in 20 countries using 15 different X.400 implementations, the public X.25 network being used for all international traffic. The work of the pilot project has led directly into a service under the COSINE Project. C. The IXI Project. In 1988 RARE drew up a plan to establish a pilot X.25 backbone network to interconnect the public packet switching networks and the private research networks across Europe. This resulted in the establishment of the International X.25 Infrastructure (IXI) by the Netherlands PTT (PTT Telecom) under contract to the CEC on behalf of COSINE. RARE provides overall direction to the project. Availability and fault statistics since July, 1990 are as follows: Table 1 July Aug Sept Oct Nov Dec Availability 95.0 99.4 99.7 94.8 97.0 98.1 The traffic growth measured over the IXI network shows a relatively small usage in the early months, reflecting the availability and reliability problems i n the initial software release. These problems were resolved in July, and the use of IXI started to increase significantly by the end of the third quarter. This trend is continuing. The traffic measured in megabytes, and in number of X.25 calls is as follows: Table 2 July Aug Sept Oct Nov Dec Traffic (in Gigabits) .2 .6 .15 1.3 3.9 5.0 Calls (in thousands)160 220 300 380 390 470 D. Other Services. S2.1 - Message handlings systems / X.400 internetworking. The purpose of the COSINE MHS service S2.1 is to ensure interworking between the national X.400 services available to research workers in the COSINE member countries and to assist with the creation of, and migration to, national X.400 services in countries where they do not exist. S2.2 - MHS gateway to the United States. The international movement of electronic mail may eventually use PTO-based ADMDs but these are only emerging slowly. During the COSINE implementation Phase it is essential to provide an electronic mail gateway service to North America to support access between the European and North American research communities. COSINE Register. This is a non-exclusive computer database used for administrative and management purposes that contains references to all organizations that have expressed an interest in the COSINE Project. The register is used as a mailing list for the announcement of events such as the COSINE User Forum, as well as notifications of Invitation to Tender. Currently, the register contains entries for some 500 organizations. Connectionless network services pilot project. A small scale pilot project to gain experience with the inter- networking of products based on ISO-IP standards. III. SPONSORSHIP OF CONFERENCES RARE sponsors an annual networkshop. In 1990, RARE and EARN jointly sponsored a conference known as the EARN/RARE Joint Networking Conference. In 1991, this became the 2nd Joint European Networking Conference organized by RARE in cooperation with EARN, EurOpen, IAB, NORDUNET, and the French Ministry of Research and Technology. IV. Liaisons with other Organizations RARE provides a user's voice on a number of European standardization and political bodies, for example European Workshop for Open Systems (EWOS), European Council of Telecommunication Users Association (ECTUA) and European Telecommunication Standards Institute (ETSI). On a broader scale, RARE represents the European participation on the Coordinating Committee for Intercontinental Research Networking (CCIRN).