EARN BOD 01 84 EARN Document Title: Minutes of Board of Directors meeting February 20, 1984 Author(s): O H Martin Date: 1984/3/14 Committee: Board of Directors Document: BOD1 84 EARN-MIN – LISTSERV@UKACRL Revision: 0 Supersedes: Status: Final Maintainer: Board of Directors Access: Board of Directors ———————————————————————— INTRODUCTION Welcome (D. Lord) On behalf of Cern D. Lord welcomed the participants to this first EARN meeting. EARN Objectives (H. Budd/IBM) Budd briefly reminded the meeting what the IBM goals were, namely to help the European Academic and Research Community establish a communications network similar to BITNET in the United States and therefore to facilitate the information flow between the two academic communications. The aim of IBM is to give the initial push and the hope is that this will allow EARN to get off the ground much faster than BITNET. IBM’s involvement in EARN is to make the Backbone Network and to bear the cost of the communications equipment. EARN Status (A. Auroux/IBM) Status of EARN Backbone: EARN phase 1 now complete: City University of New York (CUNY) <—> IBM Scientific Center (Rome) EARN phase 2, completion expected end march 84: * ROME <———> IBM Scientific Center (HAIFA): * ROME <———> CERN (Geneva): * Geneva <——-> GSI (Darmstadt): * Geneva <——-> HEC (Paris): EARN phase 3, completion expected April-May 84: * Geneva <———> Rutherford Appleton Laboratory (RAL-London): * RAL <————> University College (Dublin): * Paris <———-> Stockholm University (QZ): * ROME <———–> IBM Scientific Center (Madrid): * Geneva <———> RAL (London) *** completion date unknown ***: EARN Status by country: Switzerland: Dial-up terminal connections initially, for the Swiss Universities but a number of leased lines connection will exist before end 84. Germany: 24 Universities connected to GSI (Darmstadt). Israel: 5 Nodes planned in Haifa and Tel Aviv. France: 3 Nodes in Paris. University of Montpellier and the existing French University Network to be connected later this year. Great Britain: The RAL Node will act as a gateway to the Joint Academic Network (JANET), a private X25 network, with over 200 computers, running over leased lines. RAL will also act as a transit node to University College Dublin and from there to University of Cork. Sweden: The University of Stockholm node will also act as a gateway to the Swedish University Network (SUNET), an X25 based Network running over the Public Packet Switched Network and connecting the 60 or so computers belonging to the 6 University regions. Spain: 3 Universities (2 in Madrid, 1 in Barcelona). Most European Countries are expected to be connected before end 84, meaning that by then several hundred computers will have direct or indirect access to EARN and hence to BITNET in the US. BITNET (IRA FUCHS/CUNY) BITNET STATUS BITNET which was started in May 1981 by the City University of New York (CUNY) has now grown into a 150 Nodes Network located at some 50 sites. BITNET is an RSCS based Network providing Mail, File Transfer and interactive message facilities. Given the choice of higher level protocols which users may use, the capabilities to send and receive messages interactively is considered to be a very essential feature of such an heterogeneous network. A sizeable fraction of the computers connected to BITNET are non-IBM computers some of which running full RSCS emulation software. Please refer to Appendix A for more information on RSCS emulation software. With the existing US topology the maximum number of “hops” between farthest apart nodes is between 8 and 10. BITNET does not cater for interactive Terminal Traffic although this is possible between IBM mainframes using the Passthru Program Product also known as PVM. The use of Passthru implies the existence of either dedicated leased lines or multiport modems up to the computer the user wishes to establish a session on. The University of Maine has, however, developed code which eliminates most of the above requirements by making use of the Interactive Message facility to carry the terminal traffic over the RSCS line. With that code the only requirement is to have PVM installed on both ends. Doubts regarding this approach were expressed, as Full Screen Terminals may generate lot of traffic, which may in turn seriously impact the performance of the RSCS line as Interactive Messages are pre-empting File Transfers; also the users expectations are obviously somewhat different with respect to File Transfer or Interactive response times. Criteria for Joining BITNET: Install and pay for a leased line to a convenient node, allow at least one other leased line to another node, to allow traffic to transit at no charge; agreement to not use the network for commercial purposes. Obviously the actual commitment made by some of the major switching nodes like CUNY, University of Berkeley or Pennsylvania State University is more significant as it implies more 3705 ports, adequate amount of Spool Space, CPU cycles, etc. Statistics on the usage of the Network are collected by the individual nodes and sent to the University of YALE who has developed a data reduction program. As it standards today the average line utilisation is still fairly low (ie between 10 and 20%), other details are not known (e.g. peak values, file size, etc.). Protocols: In addition to the standard IBM facilities available under VM for Mail and File transfer, there are a number of other Mail systems in use, originating from various places (e.g. Cornell, Harvard, Stanford, etc). The recommended standard for use over BITNET is the ARPANET standard RFC822. The ‘Netdata’ format used by standard IBM facilities does not conform with the RFC 822 standard and is hence not usable over most Gateways. Columbia University has developed and is distributing the ‘MAILER’, a VM only product, which understands the RFC822 headers, the distribution lists and the topology of the network such that there is never more than one copy of a Mail File leaving a Node on a given link. BITNET Gateways BITNET <—> VNET Gateway: VNET, the internal IBM Network, with approximately 1200 nodes on it, is still growing at the rate of one node a week. The gateway software which should enable some level of communication between the BITNET and VNET user communities has been designed and written at CUNY and is now undergoing certification by IBM. H Budd pointed out that the decision of having a VNET gateway on BITNET had not been taken by IBM yet, should such a decision be taken it was however likely that there would also be a VNET gateway on EARN. The Gateway software has been tested between CUNYVM and the Cambridge Scientific Enter, the main design aspects are the follows: Gateway Nodes control the communication path between VNET and BITNET. Each Network controls access-security. No modifications required to other than the Gateway Nodes. Two level addressing, since no change required to routing tables in other than the Gateway Nodes. The communication is asymmetrical, although it does not have to be, with the VNET user having to initiate the call; this decision was taken primarily because of the different nature of the two user communities, there are thousands of students on BITNET! Prior to establishing a communication a ‘path’ or ‘virtual circuit’ has to be set-up, there are a number of commands available to set-up, one time, fixed time or permanent circuits. Finally it was also pointed out that the existing IBM nodes on BITNET (e.g. Yorktown Heights) were not using the gateway software and that access to BITNET was restricted to a very small number of users. BITNET <—> ARPANET Gateway: The University of Berkeley gateway to ARPANET is difficult to use but by the time EARN comes up, the University of Wisconsin gateway to ARPANET, which is easier to use and well documented, in addition to being only one ‘hop’ away from CUNY, is expected to be operational. Gateways to other Networks: The interface between BITNET and CSNET a network linking the Computer Science Departments of the major US Universities and a few other Universities abroad caters for File and Mail transfer but there is no provision for interactive messages because of protocol incompatibilities. A large DECNET network, called CCNET, is accessible through Columbia University same status as above with no interactive capabilities except ‘FINGER’ a QUERY type facility. There are also gateways to USENET giving access to a very large number of Unix UUCP Systems (ie over 1000), and MAILNET (same technology as PHONENET with dial-up connection). BITNET ORGANISATION AND ADMINISTRATION Organisation: One representative per site, Two meetings a year hosted by members, decisions are taken by consensus not by majority. There is an agreed set of rules (e.g. max. file size, etc) and several months notice are required should a node decide to leave BITNET, a situation which has not occurred so far. Administration: The Executive Committee is a body of six people talking together frequently in order to agree on ways to resolve problems. Systems Group, meeting twice a year, concerned with protocols, gateways, etc. User Services, meeting twice a year, concerned with Documentation, User Interfaces, Feedback from End Users, etc. Billing: There is no billing between BITNET and both CSNET and ARPANET. In the case of MAILNET, the User Community has found it so profitable to have a connection to BITNET that it has accepted to bear the cost of all communications both ways, whether this policy will be maintained when EARN is connected to BITNET is still not decided. BITNET SERVICES A VM Service Machine called ‘BITSERVE’ is providing the following services to the BITNET community: Name Server Services Directory File Server Bulletin Board (Community Bulletin Board System – CBBS) The BITSERVE machine uses VSAM to perform keyword searches in the various subdirectories and was said to be both fast and efficient. It was also indicated that answer to queries were normally sent back as Interactive Message(s) up to a predefined threshold, which when reached triggers the sending of a short message recommending the use of the Sendfile command instead. Although ‘BITSERVE’ will also run at ROME, Ira Fuchs expressed the view that the directories would preferably be kept at CUNY until such time as adequate mechanisms are implemented to allow the ROME and CUNY Service Machines to automate the exchange of information. There is for example a project to get the BITNET and CSNET Name Servers to talk together. Restrictions on Usage: The limitations on file size which were agreed when BITNET was set-up are no longer deemed adequate (ie <300MBytes between 8am and 8pm Local Time, and 3MBytes outside) and the problem will be redressed at next Director’s meeting. This problem is especially critical on the transatlantic line, hence a tentative solution to it is being developed jointly by ROME and CUNY (ie breaking large files into pieces or RSCS enhancements to support multiple Output Streams). Commercial Traffic (e.g. Administrative Use is not commercial, but fixes to Software Products cannot be distributed over BITNET). Once a new connection has been established there is no control of the actual usage which is made out of it. I Fuchs admitted that things were becoming rather more complex as the fourth generation of BITNET users was not emerging. Future Objectives Investigate the use of protocols other than RSCS/NJE (e.g. TCPIP or other Full Duplex protocols), X25 not considered at this point because of the packetisation overhead!). Provide adequate Conferencing Facilities (e.g. use of the GILT protocol). EARN EARN User Group The proposal to set up a User Group in order to co-ordinate the necessary effort to get EARN off the ground was immediately accepted by all the participants. Dr Hultsch mentioned that the SEAS organisation was willing to allocate some effort to EARN, for example as a SEAS project. Although all participants were grateful to SEAS, it was in general agreed that it was preferable not to set up any privileged relationships with other User Groups. H Budd stated that IBM would feel extremely uncomfortable about it, I Fuchs explained that BITNET had carefully avoided special relations with the SHARE organisation, even though out of convenience some meetings had taken place art the same location and time as SHARE meetings. Finally P Bryant reminded the participants that if EARN was to encompass the existing subnetworks (e.g. JANET or SUNET) then there would be a vast majority of non-IBM systems connected to what one might then call a ‘virtual network’ and hence that a close association with SEAS could do more harm than good. The following organisational structure was then adopted: EARN Association, de facto members being the Nodes with one vote at the General Assembly to be held once a year. Board of Directors (EARN-BOD) with one representative per country either elected or nominated by their country, will meet twice a year. Country representatives to say for a minimum of one year. This board will deal with: – Authorisation: – Contacts with the PTT’s: – Legal aspects: – Usage restrictions: – Gateways to other Networks: Because of the international status of CERN it was agreed to treat it as a country, and that the same rule would also apply in the future to similar organisations. The request made by IBM to be invited to the EARN-BOD meetings was accepted and it was also agreed to exchange invitations between the EARN and BITNET BOD’s. EARN Technical Meeting, with no limitation of the number of participants from a particular node or country. The criteria for membership is interest and willingness to make active contribution. The Technical Meeting which is responsible to the EARN-BOD, will also interact closely with it. The meeting will have a project structure and will initially concern itself with the following subjects: – Protocols: – Reliability: – Operations, Reporting on network load and usage, etc: – End User Facilities: – Technical problems related to gateways with other networks: Prior to being approved two aspects of this structure were discussed in detail: – With one member per country and ALL European Countries expected to join EARN the BOD may become too large, it was however agreed that it was difficult to see, at lease initially, how one could have a more restricted membership. There was also some fear that, for budgetary reasons, some Universities may not have the necessary resources to be represented at every BOD meeting, what would one then do ab out it? – Regarding the participation to Technical Meetings it was finally agreed that there was no good reason to restrict it to one person per node. It was also pointed out that because of the Network which was being created very useful co-operation and contributions could be made even without attending the meetings. At that point the participants were reminded that EARN will only be concerned with the Reliability of the Network and NOT with SECURITY Aspects. Responsibility for that remains with the individual nodes. It was also agreed that the NON-Secure side of EARN, hence the need to Encrypt Files whenever secure file transmission is a requirement, had to be very clearly stated to ALL EARN Users at the time of joining the Network. User Responsibilities and Procedures for Joining EARN BITNET: A new BITNET member must contact the nearest Node, make the necessary arrangements with the telephone company and sign the ‘BITNET agreement’. In practice the BITNET BOD is informed beforehand, although it does not have to be, unless there are doubts as to whether the applicant qualifies for BITNET membership or not. At present full membership is restricted to academic institutions, however the National Laboratories are able to join as they are run by Universities. In general this is not the case in Europe, hence a potential problem area when European organisations like CERN wish to be considered as having full BITNET membership. Another rule is that every document passing through BITNET must benefit to a BITNET member, meaning that traffic via Gateways not originating from or going to a BITNET node is hence prohibited. The issue of what would be the status of the various nodes of the EARN sub-networks was then raised (ie JANET or SUNET). In addition to the ‘No Commercial Use’ rule mentioned earlier two categories of BITNET members have been defined: – Fully fledged members, also called Class A Users – Non-fully fledged members, also called Class B Users The difference between class A and B users it that class B users are only allowed to talk with class A users. There is, however, no mechanism to date to prevent class B users of talking together. Class B users are non-transit nodes (e.g. IBM Research at Yorktown Heights was quoted as a typical class B user). P Bryant asked for a copy of the BITNET agreement for examination by the PTT’s. EARN: The situation in Europe is somewhat different because of the many national PTT’s involved and the rate at which the network is expected to grow during its first year of existence. The presence of /Academic Networks in several countries makes the situation even more difficult to manage as demands for connections to JANET for example will continue to be handled according to JANET rules, also connected to JANET are a number of Local Area Networks which are outside the control of the JANET Board. The following procedure for joining EARN was suggested: – Contact Country BOD: – Find appropriate Node: – Sign the ‘EARN Statement of Conditions’ or Make a Statement: that one has seen and understood the EARN rules as laid down in this document (see also below) – Interface to EARN: – Inform EARN-BOD via Country representative: Criteria for joining EARN (tentative): – All academic institutions (ie Universities or similar): – Public funded and/or controlled non-profit research organisations: – Other academic/research non-profit organisations: Note that Libraries, for example, would not quality. At line subscription time the PTT’s usually ask what the intended use of the line is, hence it was felt very important that the policy for using the line as well as the criteria for joining EARN, be formulated in the ‘EARN statement of conditions’ and presented to the PTT’s beforehand. Equally important is to define how does one tackle the problem of users transgressing the rules. If the rules for using the network are ill defined and if the criteria for joining are too loose it was also felt that there was a genuine risk of EARN falling apart because of the PTT’s not renewing some licenses after the first year of operation or introducing volume charging. D Jennings and D Lord agreed to draft the ‘EARN Statement of Conditions’ for presentation at next BOD meeting. In the meantime I Fuchs will make the ‘BITNET agreement’ available and P Bryant will check whether the JANET regulations seem to be ‘a priori’ compatible with BITNET or not, the hope is that they are more restrictive and hence that it should not be a problem for a JANET Node to become a Node on BITNET or EARN as well. It is not clear yet, and it also depends on the gateway implementations, how JANET or SUNET Nodes will appear on EARN and BITNET and which status they will have (ie Class A or B vis a vis BITNET, the purpose of EARN would obviously be defeated if the EARN rules were watch that for example JANET and SUNET users were not allowed to use EARN in order to communicate together). Relationships with other Networks: There is a general problem of BITNET or EARNET being used as Pass-through networks, however as long as lines are used well below their capacity, this was not considered to be a very critical issue. Bilateral agreements being easier to reach, I Fuchs suggested to negotiate the access to CSNET over BITNET with CUNY and the University of Wisconsin. Relationships with other Groups: It was agreed not to establish special relations with existing user groups like SEAS, DECUS, etc. Operational Problems EARN is expected to be operational 24 hours a day, 7 days a weeks which is obviously, an objective, difficult to reach. To minimise the inconvenience of interrupts to the service, it is recommended to make schedules known to other sites (e.g. Preventive Maintenance or other planned outages), this is especially important for the major transit nodes like ROME or CUNY. D Lord mentioned that each year CERN was officially closed during the week between Christmas and the New Year. Node Naming I Fuchs recommended the use of good mnemonics with some mention of the Name of the Operating System. Otherwise Names have to be unique and should preferably not contain special characters, if a site is planning to have several nodes connected it is a good practice to ‘number’ the names. ‘ALIAS’ code is available under VM from University of Maine for those sites who cannot or do not want to change the name under which they are known locally, but wish to appear on the network under a different name. Germany has already adopted its naming structure and will stick to it. Given the facilities generally available for using nicknames instead of the actual names, this problem is of no particular importance. A more serious problem is the one of addressing users over the gateways to other Network, and is a subject of further discussion between P Bryant, D Jennings, P Hebgen and O Martin. Network changes are accumulated at CUNY and are available via BITSERVE commands,m New Routes are distributed to everyone in BITNET and I Fuchs kindly proposed to continue doing it and to extend the distribution to All EARN nodes. Name and File Servers: Most of the request to the Name Servers being expected to be at the country level, it was suggested to start with a two level Name Server structure (ie Network wide at CUNY for BITNET and EARN, and Country wide in Europe). Future Meeting: Next meeting will take place in Paris on May 14th. APPENDIX A (RSCS EMULATION SOFTWARE) The VAX/VMS Software is now being marketed by: Joiner Associates Inc 1124 Edgehill Drive Madison Wisconsin 53705 Tel: (608)-238-8134 An implementation on PRIME computers is also believed to be available from the same company. The VAX/UNIX Software is available from Pennsylvania State University Computer Science Department, Professor Robert OWENS. Tel: (814-865-1131). APPENDIX B (LIST OF PARTICIPANTS) A Auroux – IBM Europe H Budd – IBM Europe P Rodier – IBM France B Kirchner – IBM Germany P O’Connor – IBM Ireland A Fusi – IBM Italy M Rimon – IBM Israel J Becerril – IBM Spain C Heritier – IBM Switzerland C Setford – IBM United Kingdom J Ippolito – France, CNUSC F Bush – Germany, GSI Darmstadt L Hultsch – Germany, GSI Darmstadt R Wolf – Germany, GSI Darmstadt P Hebgen – Germany, Heidelberg University D Jennings – Ireland, University College Dublin A Cohen – Israel, Inter University Computer Center M Sommani – Italy, CNUCE L Mate – Spain, Universidad Politecnica de Madrid S Eriksson – Sweden, Stockholm University Computer Center (QZ) T Hatt – Switzerland, Zurich University A Kundig – Switzerland, ETH Zurich P Bryant – UK, Rutherford Appleton Laboratory (RAL) I Fuchs – United States, City University of New York (CUNY) D Lord – CERN O Martin – CERN Share this:FacebookTwitterReddit