Paul Bryant's Networking Correspondence
I have now investigated the dial up problems you brought to my attention. I will be discussing the questions at NDM on Wednesday, you may like to attend for that item.
We have now found that the Texas terminals which exhibit the problem of being unable to use the new equipment can be very easily modified to overcome the problem. I have spoken to James Hutton and he is happy for us to modify his terminals and he then has no objections to the removal of the BT dial up equipment. I suggest we do this. Andy can do it.
It is worth adding that we have recommended that acoustic couplers should not be purchased for a long time and I was a little disappointed to see that we had so many. We should reiterate this recommendation especially as BT liberalization has given us an abundance of cheap choices.
I have now had a look at the memo from Eric. It worries me. it has been well known to Eric for 6 months that the dial up project has been in progress. Informatics have been aware for some months that the new dial up equipment has been available for testing. Informatics attend NDM where dial up plans have been discussed from time to time. It is disappointing that a day or so before we intend to de commission the BT equipment that we learn of Informatics problems. It must be added that we are unable to do exhaustive tests on all pieces of equipment due to our limited expertise and must rely on members of other divisions noting our plans and taking steps to ensure that they do not effect them adversely.
Eric makes a number of claims which I find puzzling.
First. Peter assured me that at no time has he discussed the provision of the 2400 bps service via PACX. I now find that Eric sent Peter a mail note via the IBM. Since Peter does not often use the machine he only found it a day or so ago. I find it quite remarkable that Eric on not receiving a reply did not follow the matter up. As an aside, this is the result of the pathetic mail system on the IBM that it is incapable of allowing mail to be forwarded automatically. With such a mail system it is unwise to assume that because someone has an account that they use it frequently, I only log into the IBM infrequently. Anyway, I am not happy to provide 2400 dial up via PACX as it goes against the policy of moving slowly towards terminal services via the PADs, and no extension of the PACX service and the eventual decline in PACX when the network services are improved.
Second. It is claimed that the 'network removes control characters'. There once was a problem with XON and XOFF with CAMTEC PADs but I am now assured that in NATIVE mode no characters are removed. This is not necessarily true of other PADs but I do not think that is relevant.
Third. I am unaware that the quality of the UNIX triple X service was so bad that it could not be used for dial in traffic via PADs. I have no UNIX experts and expected Informatics to run any necessary tests since they attended NDM and knew what was being planned.
Fourth. I can see no reason why the reverse PAD route cannot be used although it is certainly not quite so convenient.
If Informatics demand that direct services are required then I see four options:-
I favour version 1 if Informatics cannot get over their problems.
I must again state that this problem has come from the deliberations of NDM not being taken seriously. I would like to draw to your attention for the purposes of your discussions with Informatics that the project plan for the dial up project has been available for many months and was approved by Group Leaders at the time Eric was a member of that meeting. The project has had almost no changes in detail. The progress of the project has been discussed at many NDM meetings to which any one is welcome to attend and which is normally attended by Informatics.
Lastly I am confident that the project has been run properly and all reasonable steps had been taken to test the equipment prior to the launch announcement.
Could I ask you as the recipient of the memo from Informatics to clarify and prove the claims they have made and if substantiated to agree a course of action through NDM.
I apologize that you seem to be the innocent party in the middle of this match.
Thank you for the project plan for the 4705. May I distribute this to NDM for their meeting tomorrow Wednesday?
You may well like to attend the meeting to help us.
There is one point that worries me. Although you show the set up of the Memorex 1270s you do not show the corresponding set up of the 4705. There is a comment that suggests that all the lines will be carried across but some of the existing ones appear to be of a temporary nature and it may be sensible to take the opportunity to make a few changes.
A particular change I am interested in is whether the Full Screen Cifer connections should be maintained rather than insisting that all such traffic goes across X25. The point is that the number of ports is small and I am not sure that they are used very much. I have an ulterior motive in wanting this change in that CERN are interested in the code to interface to their X25 COMPRO code. I believe that Peter's code is technically a better solution. If we were to show signs of phasing out the direct code this would strengthen my hand with CERN.
I am also still a little worried over all the curiosities we have and would like to have a careful look at any rationalizations that may be possible. In particular I do not like that disgusting affair they call passthrough which nags at my aesthetic senses- there must be a better way.
This was an international meeting called to consider the COMICS mail report. I have many copies of this report and anyone interested is strongly advised to read it. In brief is states that now there is an international mail standard, called X400 or MHS, the HEP community would do well to migrate to it as soon as possible. It recognizes that this will take some time and so gateways between X400 and various other mail systems will have to exist for some time. Already there are some X400 implementations that are available and being looked at by CERN.
The meeting was propped up by a well known gang of Les Robertson(CERN), Ulf Beyshlag(CERN), Jack Prevost(Paris), Jim Craigy, Tom Kokott, G Henken, Tommy Erikson and Enzo Valenti - a good cross section.
The plan was unfolded. Initially it is envisaged that X400 and all other systems would interface into some vast RFC822 UNIX based mail switch. As time went on the various mail systems would move to gatewaying into X400 and so X400 would become the centre of the mail world rather than just another mail system. Initially it seemed sensible for the initial system to use RFC822 protocol as there was a lot of it. This is a gross oversimplification. Read COMICS to get the true flavour.
The meeting was unanimous is agreeing that X400 was the way to go and all the discussion was on points of detail.
There was some opposition to HEP forming a mail domain which would overlay the mail domains elsewhere. I felt the topic needed more study. If users found them selves in a multi name domain it could be very confusing.
The current products of most interest are the EAN ones from British Columbia. These work quite well. It seems several manufacturers are interested in taking the products. The IBM product, as usual, proved the biggest stumbling block. The product from Waterloo will have X400 over RSCS instead of X25. I will not go into the gymnastics for going though various gateways to get this turned into X400 over X25.
In a round the table survey we learn that Siemens are taking EAN. ICL have some product which is not portable (what machine?). UNINET and possible DFN are taking EAN. NORSK are debating who should do it- to be translated- can they get out of doing it! DEC are working on X400 for the message router project whatever that is. It is expected by the end of the year but may be a closed system. ESPRIT want it under UNIX for the ROSE project.
Most of the PTTs will provide X400 between their mail boxes. Since most of these seem to be based on DIACOM (including BT with TELECOM GOLD) we may expect them all to say 'go' when the product is finished.
There was a cantor round the academic networks which. France expect to have X400 in their so far virtual network. This is likely to be provided by Bull. IN2P3 will use X400. Italy still stuck on DECNET but OSIRIDA will probably follow DFN. OSIRIDA seems to be currently defined in a single slim line document. However it will use ITAPAC which now seems in good shape. JANET, of course, has lots of plans foe X400 under active development. Denmark has nothing going on. Norway intends to adopt EAN. Sweden will use X400 and will have X400 in PORTACOM by the end of the year.
Joking aside the meeting did demonstrate considerable support for COMICS and X400. Nobody seriously expected it not to happen. The question was when and on the detail of the transition.
I did raise the question of how PROFS could use X400 but CERN had not considered the problem but I warned them that this was an important issue.
These meeting have adopted a certain style and under the able chairmanship of James Hutton continued in the style of constructive chaos. The meetings are characterized by the stream of members coming and going at intervals, presumably to go and search for a lost omega particle or two. This leads to the agendas being modified to reflect the attendance at any point in time. The minutes should prove interesting and a challenge of some magnitude to the secretary.
I will only mention some of the more interesting discussions.
GIFT had some strong words from Francoir Flukiger who was very worried over the problems at Rutherford. It is certain that the conduct of this project has not enhanced our reputation. To be fair the need for a P end had not been seen as all that important. Also, to be fair, Willy Black has done a tremendous amount of work and Sue Weston is becoming very useful on the project. None the less flack will fly! Still- we are doing better that UNINET who have not started. The DECNET part seems in good shape.
EURONETWORKSHOP was enthusiastically welcomed. The idea that it should lead to some sort of Europe wide network team to organize things seems well accepted. It is hoped that the EEC or COST11 may cough up. The relevance to HEP was talked about and it seems certain that HEP will have a lot of influence on any outcomes.
Richard Mount gave an entertaining but worrying review of the HEP networks in the USA. There is a lot of protocol confusion. He gave some fascinating cost studies which I did not manage to write down. It seems likely that there will be a leased line across the Atlantic for LEP. Seems the LEP VAX machines in the states will run Coloured Books over the line to CERN. It is expected to be operational by end 1985. It will cost a fortune.
EARN came up. There is not over much enthusiasm for it. I learned that Farber, Landweber and Fouchs are cooking up some migration from EARN protocols using TCP/IP over X25 level 2. They will be using PC/ATs as switches and will get into the IBMs via Ethernet and the DACU. Seems a pity they are not going ISO as what they are doing does not seem to help the EARN transition to ISO.
SCIENCENET in the States looks like being based on TCP/IP. It also seems that the DOD is supporting Transport Class 4 as a migration from TCP/IP. Again, seems a pity.
Optical discs were discussed (don't know what this has to do with networks). The opinion is that the technology is still not ready but it would be useful to have some pilot projects to look at the various products that do not quite work.
The ESONI LAN project is now more or less dead. That is a pity since a lot of people had spent a lot of time on it and some thick brown stuff flew in the direction of the organizer. The move to ISO for data collection and distributed services seems to be very slow as there is still a lack of protocols. In the mean time work based on TCP/IP is going ahead.
Naming came up for the usual mega discussion with the usual mini results. It was concluded that Euronetworkshop should initiate a Europe wide discussion on the subject. This might lead to a Europe wide NRS.
A lot of other topics were considered of less interest to me. This was a very lively and useful meeting. A lot of good work was done outside the meeting while searching those elusive particles.
Dear John,
I guess you know about the European Academic Research Network (EARN) which is a set of ~100 computers linked together across Europe with the IBM RSCS protocols. As a result of allowing the network to exist CEPT is insisting that it should migrate to ISO protocols and use public networks in the next 4 years. Although the EARN management is independent of IBM the network is financed for the next few years by the company.
It so happens that I am the technical coordinator of EARN and on my shoulders rests the problem of how this migration should take place. IBM has been most supportive and are providing facilities for my activities. The first of these activities is a meeting with IBM experts and EARN experts in La Gaude to examine all the options and come up with some recommendations of how to proceed. I have found that the level of understanding of the options and, indeed, of ISO within EARN and to some extent IBM is a little depressing. Moreover the enthusiasm for the studies is less than I would like. Hence it looks like much of the work is going to land on my shoulders.
I guess you can put names to the options as well as I can. A short list is:-
Coloured books followed by migration with UK. Probably unacceptable to Europe. RSCS over X25 being developed by CUNY. Could be a useful migration route but where are the subsequent stages. Use of IBM ISO products. These are part of the SNA stable. There seems to be some understandable reluctance to embrace SNA in EARN probable due to the costs, particularly manpower, involved. In addition, it is unclear what IBM intend with their products. Also X400 and FTAM do not seem to be on their list yet.
There are other question as to whether options based on a Series 1, 3705 with or without COMPRO software are best.
There are rumours of other developments but they are difficult to track down.
A strong possibility is the FTAM product you are proposing to produce. This would certainly meet a lot of the requirements. The important missing component would be X400 since mail is one of the most important uses of EARN which currently uses RFC822 over RSCS. Regretfully the X400 product being produced at Waterloo is designed to work over RSCS.
I would be most grateful is you could give me as much detail of your product as possible to help us evaluate it against the other options. My time scale is that I will be spending a week in La Gaude with the experts in June. As well as technical details it would be useful to know about availability dates and costs. I would also welcome any other advice you might have. Please bear in mind that my experts know an awful lot about IBM stuff and not much about protocols.
Best wishes,
Paul Bryant.
* Uses IBM RSCS protocol * Machines linked using leased lines * Provides file transfer and mail * getting on for 100 machines * Same scheme as BITNET with 500 machines * Two lines to USA
* EARN is an ASSOCIATION registered in France * Run by Board of Directors - one from each country. Meets a few times a year * International lines paid for by IBM for 4 years * EARN independent of IBM
EARN only allowed to exist if:- * Migrates to ISO protocols * Will use public networks CEPT recommends * A volume tariff is applied - many countries apply a zero tariff
* EARN has a technical group led by a Technical Coordinator * Meets a few times a year * Small subgroup is studying migration options with IBM experts. Expects to make recommendations in June * Will be meeting in La Gaude
* IBM X25-transport & session via 3705 part of SNA * RSCS over X25 from CUNY * Coloured Books Rutherford style via 3705 and COMPRO or bisync and stunt box * Coloured Books Salford style via Series 1 * FTAM from Salford via Series 1
It is unclear whether EARN will attempt to continue with leased lines after migration No studies have yet been made for converting to use X25 switches- current topology not suitable
No obvious migration path
The nature of IBM ISO support unclear.
Will it be suitable for academic use?
No X400 product or CCITT 84 in sight
I go along with most of the points you make. To avoid confusion I will comment on all the points.
First - The principle problem is lack of effort. As you know I had hoped to increase my staff on IBM networking by one just to overcome current problems, this was not to be. Also I only have and can afford 6 man months of effort on the gateway which will certainly be only enough to get it going and not to polish it.
I attach a copy of the EARN Charter.
Yes there should be a user group and I hint at that in section 5. I was none to specific as I am very unsure how it should operate. A fundamental question is whether it should be EARN members i.e the sites or EARN users. I am very reluctant to have two separate meetings for both groups. The users would be more fruitful but it is the sites who sign the Charter. It may be that the sites should be invited to send users. A further point is that effort is needed to run a meeting and there is no effort. In principle User Support should run it and although I have not approached then I am sure they will also plead poverty. I am sure an offer from JNT to organize it would be gratefully accepted! I would envisage a meeting once or twice a year. My experience of user meetings is that there is a great reluctance to attract actual users rather than network junkies. I would point to the almost lack of users at the JANET regional meeting and the fact that the GEC ICF user groups tend to be packed out with computer managers. However, we must try and be seen to be taking account of the users. I have amended the text accordingly. In any case there is no point in a user meeting until there are a reasonable number of users.
For better or for worse it is almost impossible to give users not registered on the IBM the same service as ones registered. This will be possible with mail but not for file transfer. It is not possible to stop users of the IBM using EARN 'as it was intended' rather than via a FTP to EARN gateway. They will also have access to all the other curious features of EARN which are difficult and possible counter productive to gateway. For example - EARN users now have a CHAT facility to allow terminal to terminal traffic. There is no comparable service on JANET and in fact it may well be undesirable to allow such a service without the relevant protocols being sorted out. Again to provide these extra services effort is required that does not exist.
Yes NRS names will be used. However the only names to be put up will be those of members of EARN. It would clearly be ridiculous not to put such names up as it would make is very difficult to get EARN JANET traffic without doing so. Thus it is certainly not the intention to put up all Grey Book mail services registered in the NRS. For traffic from JANET to EARN all EARN BITNET names will be available. These will be EARN BITNET registered names. There is no intention to resister EARN BITNET names in the NRS as they are a separate naming domain (you may wish to persuade me different - if you do we have an interesting problem of how we keep the NRS up to date - manpower again). I do intend to register EARN and BITNET as top level names on a level with UK (again this is something to discuss). Thus the mail name of someone in EARN will be, say, FRED@DEARN.EARN. You will note that there will regretfully be no way of preventing traffic from anywhere in JANET from going through the gateway.
How much functionality we can put into the MAIL gateway depends on effort. I would hope to put in the VIA and RECEIVED FROM fields and this looks very possible. I do not understand your comment of error reports going to users rather than administrations. This seems to be a function of the mail implementations on various JANET machines. How could I make an error report to an administration when I don't know their address? However I should add that there are severe problems in providing error reports as the MAILER we are adopting does not expect to be driven as a gateway. For example the only way the mailer can find out if an address is valid it to fire off an FTP. If this is rejected then FTP will correctly report the fact but by that time the MAILER has lost all track of the mail and FTP will not return the document for the MAILER to work on. This would require a lot of effort to sort out properly.
We are still working on the details of the gateway and would certainly like comment from users although I would not like work to be held up pending a user group meeting. In fact it is likely that there will be very few users by the time work has to start.
On FTP the situation is much clearer. We are doing nothing as there is no effort. The scheme to be used will be the scheme which has been used for a long time for sending file traffic to and from CERN. Thus to FTP a file you send it to device LP at UK.AC.RL.IB with user id FRED.DEARN say. Coming the other way we have a severe problem that RSCS has not got enough address information to carry it through JANET. This is overcome by putting in the so called *FILE records in the text which can be dealt with in the RL IBM to generate a suitable FTP. This is very nasty but there is not any sensible alternative other than having vast tables in the IBM of everyone who wants to receive files, even then there would be severe restrictions on what one could do.
I would rather be non specific over the time scale for migration to ISO as the time scale depends on when products will be available, when other networks migrate and what the attitude of CEPT turns out to be. With tongue in cheek I could ask you when JANET intends to migrate to ISO and I guess you can do little better than 'the next few years' even though your thinking is way in advance of anything EARN has done.
I must stick by 'IBM playing no part in the EARN management'. Yes, they do contribute technical effort and resources. But it is up to the EARN Board of Directors to accept or reject the offers. In addition, the BOD is willing, able and anxious to receive help from other manufacturers. IBM is in no way constraining the options available except to the extent that their gifts have restricted capability.
I am happy with your rephrasing of the undermining comments which are less emotive!
I do not really understand your comment on Clause 3. I am happy with your re phrasing but do not see any difference in meaning.
In Clause 5 the benefits of membership of EARN are to participate in its running. They can elect the Board of Director member and through him can influence EARN. They can also decide within the terms of the Charter how EARN should be organized within the UK. I was trying to separate the direct benefits of being able to use it and the benefits of being able to help to run it. However, I have removed the word.
On clause 6 you are quite right, the development possibilities were not detailed. I have amended the title accordingly.
Yes the loss of services in the gateway are a problem. I am a little reluctant to offer services in JANET which are not commonly available in JANET. There are many such services provided by RSCS. It is true that such services could be provided by allowing users to log into the IBM. This means that accounts would need to be set up and this means effort. We already do it to some extent in that there is a user id available that allows one to look at the job Q on another machine. This would certainly exist for EARN users. This could be extended. I am unhappy to introduce services which demand more extensive use of the IBM and the introduction of services which have no suitable protocols in JANET. This is clearly an area where a user group could advise.
Certainly the biggest problem is going to be the lack of effort.
Thank you for your comments and I attach a modified copy of the paper.
A meeting of representatives from the European academic network communities was held in Luxembourg from 13 to 15 May 1985. The meeting, was sponsored by:-
COST - Cooperation European dans la Domaine de la Recherche Scientifique st Technique.
ECFA - European Committee for future Accelerators (which co-ordinates High Energy Physics research in Europe).
ESF - European Science foundation.
The meeting decide to set up a European association to coordinate and foster the development of networking within Europe for the benefit of the academic and research community.
Networking facilities have been introduced into most academic and research institutes in Europe. A variety of technologies have been used depending of the manufacturers involved and the particular circumstances at any site or within any country. This has led to difficulties in communicating between people from various countries and often within countries. Recognizing these problems, many countries have initiated national programs to harmonize the activities within their borders.
Professor Carl Zander from the Hans Mitner Institute in Berlin recognized, in setting up a national German network, that there were requirements for communications traffic between countries and in particular between the European ones. He therefore initiated a series of meetings in late 1983 and early 1984 to attempt to harmonize the protocols planned to used in the various countries. This work was generously supported by the Information Technologies and Telecommunications Task Force of DG3 of the European Commission. This work was based on the emerging protocol standards being developed by the International Standards Organization. There was confidence that good communications facilities between the large variety of equipment already installed could only be achieved by the use of a set of non proprietary, standard and adequate protocols implementations. Such a set was only likely to be derived via ISO. The result of this activity was a set of documents which defined interworking subsets and implementations recommendations for many of the ISO standards or draft standards. These documents have been used extensively in some ESPRIT activities and within several of the national academic network groups.
In late 1984 advice was sought as to whether it was appropriate to hold a 'workshop' to consider further harmonization and collaboration. The idea was enthusiastically welcomed and an ad hoc organizing committee started to plan the venture. Political sponsorship was sought and given. A sponsors committee was formed to ensure that the activity reflected their wishes. The terms of reference of the Sponsors committee are annex 1. The organizers met and decided that the workshop should be limited to 60 or 70 people to ensure good discussion. Attendance was limited to representatives from countries and from the sponsors. To this end a representative was selected in each country who selected the rest of the national delegation. The size of a delegation reflected the population of the country.
Luxembourg hosted the workshop as the Commission generously donated meeting rooms and organization resources. The location also reflected the pan European nature of the workshop.
The program reflected the desire to concentrate on the exploitation of existing and emerging technologies for the benefit of users rather than to discuss research topics.
The workshop was opened by Professor Carl Zander who stressed the importance of achieving a good network infrastructure for the European community. He outlined how he thought this could be achieved. This was a very appropriate time to be starting an initiative and the goals were within reach. His address is reproduced in annex 2.
In the evening of May 14 a meeting was held between a representative from each country and some of the sponsors. The meeting proposed that an 'association' should be set up to foster European academic networking. There were strong indications that funds could be found from one or more of the European institutions to support such a venture. In the main these funds would be needed for travel, meetings and perhaps a small permanent secretariat. The main purposes were for cooperation, collaboration and harmonization. Pending the setting up of an association it was agreed to set up initiatives in eight priority areas. These are:-
Message handling X25 84 Collection and dissemination of information File transfer Operation of networks Full screen terminals Administration and forward planning Liaison with CEPT.
The minutes of this meeting are in annex 3
The proposals were put to the workshop in the concluding session by Professor Zander and met with unanimous and enthusiastic acceptance. A record of the concluding session is in annex 4.
After the workshop the organizing committee met and considered how the eight priority areas should be progressed. A record of their discussing is in annex 5.
A number of presentations were given during the workshop which defined the state of networking and explored a large number of options for the community. The details of these will appear in proceedings. A summary of each talk is given in annex 6.
An address list of the participants is given in annex 7.
When DFN started in 1981/2 it was realized that we must not make the mistake of not being an 'open' network. The existing computer networks needed harmonization to allow the transfer of information around Europe. We had the standards but they needed interpretation. We also wanted to eliminate parallel developments and save intellectual effort. Therefore we needed to make selections from the options in the standards. As the standards were not fully available the work had to be often based on drafts. As suppliers may anticipate how a standards will be used harmonization must be early and to a common set of standards. The Exercise was called 'European Harmonization Activity' (EHA).
This workshop supports this idea. The key role is to allow interconnection across Europe avoiding gateways and overcoming the barriers in networks but supporting human effort. What Europe needs is convergence to common information standards.
EHA was supported by representatives of National Programs from:-
Denmark France Germany F.R. (DFN) Greece Ireland Italy (OSIRIDE) The Netherlands UK (ALVEY etc.) CERN
There have been six meetings so far:-
20. Oct. 83 Strategies; Zander (D)
14./15. Nov. 83 Network-Layer; Bartlett (U.K.)
19./20. Dec. 83 Transport-Layer; Lenzini (I)
12./13. Jan. 84 Terminal-Access; Beyschlag (CERN)
23./24. Jan. 84 LAN-Operations; Jacobsen (DK)
13./14. Feb. 84 Session-Layer; Michiels (NL)
27./28. Feb. 84 Terminal-Access; Beyschlag (CERN)
These meetings resulted in the 'Common use of OSI Standards' (COS) documents. These recommendation documents are structured as follows:-
- specification of the choices made within the standard - rationales and motivation for these choices - implementation guidelines.
MHS and FTAM have not yet been studied.
We have to look at the environment and the goals. Now classical networks such as the grandfather of them all, ARPA, are not suitable. It is very difficult in EEC countries but we must do our best to converge in 5 to 8 years.
The future in the next decade is much more challenging with broad band communications. has speeds of up to 48K.
Present:- A Patel Ireland M Walsh Ireland M Wells UK F Ros Spain K Neggers Netherlands B Mahon CEC J Harms Suisse A Kundig Suisse M Paul Austria M Hind CERN B Carpenter CERN F Fluckiger CERN E Valente Italy J Prevost France U Beyschlag CERN P Villemoes Denmark B Carlson Sweden P Kirstein UK B Butscher Germany J Hutton UK P Van Binst Belgium K Ullmann Germany N Newman CEC J Renuart CEC K Zander Germany P Linington Chairman P Bryant Secretary
J Renuart thanked Professor Zander for originating the harmonization activities which had led to this meeting. This meeting has been supported at the last COST 11 ter meeting. The decision to have it in Luxembourg provided the right environment to get the right people together.
IES requires networking to increase the cooperation between the ESPRIT parties. There is the technology to do this but it can benefit from the experience of the national programs. If national programs can cooperate there can be benefits. The operational infrastructure can be assembled quickly. One good reason why funding may be available is that it may be a good vehicle to produce a critical mass of users pushing for standardization, this would be of use to IES. A second reason is that there is a lot of SPAG activity in CEN/CENELEC but there is a lack of strong user community which can cooperate with the standards bodies. Thus there is a need to create and support an initiative as a way of getting the academics to speak with one voice.
Professor Zander hoped that when he started the harmonization activity in 1983 that it would lead to a European academy of networking but for various reasons the time was not right. The foundation of EARN had caused some uneasiness and had led to the idea of preparing a framework for academic networking supported by the CEC. In addition there was a lot of other things happening in ESPRIT, RACE, CEN/CENELEC, the 'big twelve' and these activities were missing the academics. European academics have a huge potential that must be used in the race with the USA and Japan. Academics are promoters of ideas. The time is now ripe to found an association for academic networking. In it they would stay together, share experience, share risks, share costs and make themselves more effective. There is a large responsibility to provide the academics with infrastructure and the first stem is information and communication. They must stick and work together to accelerate activity. Together they will have power compared with manufacturers. The industrial research laboratories should not be excluded. Also the Scandinavians should be included. These points may cause problems which must be overridden. Europe needs academic networking in the race between the continents. European sciences, such as HEP, astronomy and so on, need an infrastructure. It is natural to came together in an association, the details are up to the academics but they must act.
P Linington asked that the meeting concentrate on principles and scope, not details.
Professor Zander outlines the goals as:-
To form a powerful cooperative association. To form a critical networking mass. To be a strong counterpart to the EARN association. To harmonize networking. To provide matched applications and message handling. To ensure convergence to standards. To exchange experience. To provide test beds with pilot projects. To ensure feed back of experience to ensure effectiveness of Europe. To initiate new OSI standards where needed. To stimulate powerful new facilities for example to super computers. To stimulate and motivate action.
M Hine stated three topics for work which must not be confused:-
To create an OSI network. Migrate current networks to OSI. The interconnection of non OSI networks.
The last item is pro OSI in that it is necessary to obtain a universal community and this must initially be done with the tools available. Having created the interconnections a further OSI network can be created in a clear way.
J Renuart supported this view which was in line with the IES need for an infrastructure now. If the policy is to wait for ISO products in some years then it would be too late and so an infrastructure has to be created with what is available together with plans for migration.
M Wells supported this comment and added that a fossilized network was no good but on the other hand the users wanted a reliable service which meant the service networks could not be used for development. He expected the transition to ISO to take between 5 and 15 years from the present inadequate ones to full OSI networks. The long term nature of the activity must not prevent it.
M Wells stated that the networks should be for all academics and not subsets such as HEP. The use of networks by the arts is likely to increase as they become aware of their potential. Arts workers have the potential to generate very large traffic requirements.
K Neggers has been impressed when in the USA by the meetings between the computer centre directors where many topics are considered. In Europe we miss this academic communication. Should the aim of the association be service or advice. His impression was that the CEC wanted service. It will be difficult to match the things needed.
J Renuart thought that the creation of networking not OSI to be the principle interest followed by research and standards. the requirement is to open communications to the whole community and this may be as a result of some standards activities such as SPAG.
Professor Zander noted a problem in Germany where several million DM had been allocated for personal computers. These need to be embedded in networks. These face us with a complicated system and we need to exchange experience and ideas. The use of networks is likely to increase quickly in non science areas and they will want more services. Computing overlaps all areas. We must decide what we are aiming at, now, 5 years or 10 years and kilobytes or gigobytes. Here we have the experience, requirements and standards and extend these back into CEN/CENELEC and other standards bodies. The average life of equipment is 3 or 4 years yet it takes 12 years for standards to be developed. These problems need to be solved in the framework of the community.
P Kirstein thought that small two or three year projects should be started. It would take about three years to get interconnections and it should start with a small number of countries.
B Carpenter thought that users should be sought who have the expertise to express their needs in a way the providers can understand.
J Renuart did not want the issues confusing and wanted to address the aims of the association.
M Wells stated that networks were facts and that people were here as they managed these networks and all had limited interconnections. We must agree at manager level that we want to interconnect effectively and the technical people will have to define how this can be done. As a manager he could go back to the UK Computer Board with a proposal to do something and he thought that there would be more impact if everyone did this. In fact the activity has to be stage managed across Europe.
J Prevost while agreeing that the first priority was the exploitation of the currently emerging networks thought that there may be a tendency to forget the importance of projects to provide the networking of the future.
P Linington saw the object as providing a network infrastructure for the research community both academic and industrial. This should include CERN. Preparation work on standards was needed but only so far as required by the infrastructure and the support of computer science was not an objective. ESPRIT interest was that they needed the infrastructure quickly.
J Renuart noted that ESPRIT now had 110 projects soon to increase to 180 and there were 600 to 700 users. This is a relatively small community and that is why it is better to combine it with the academic one. The first step should be taken on interconnection and then further steps can be taken for a new infrastructure.
J Harms pointed out that there was a wide difference between countries as some already had networks and others did not. The mechanisms for the association must be built with care and ensure that countries all agreed to give and take. It was also important to cooperate with industry.
In a discussion on the geographical region to be covered the question of East Europe was raised. M Hine thought that it would be a mistake to introduce too many incompatible people at an early stage. We already have many problems with the very different networking activities in countries. There may well be political difficulties with some countries. It was felt best to leave this question until an actual case arose.
B Carpenter was worried over bringing in industry at the beginning. Whist there is a common interest, industry do have problems of contracts and secrecy which could upset the start of a delicate operation. He wanted to see the academics reaching agreements first on how industry is brought in.
J Renuart said that the experience from ESPRIT was that universities and industry need time to adjust but that eventually it worked well. He reiterated that the aim of the CEC was a high quality network for ESPRIT. He agreed that the academics had to do it themselves to be successful and if industry is in at the start they could be disruptive.
P Linington agreed and was not prepared to take in industry if it would be disruptive to current networks.
P Kirstein has seen industry using CSNET and ARPA without much disruption. He noted that industry took precautions in only adopting specific services.
J Prevost saw collaboration with industry as important as it was the only way academic work will provide benefits outside universities. But he agreed that it would be difficult to incorporate industry at the start. We could allow industry collaboration on a project by project basis.
F Fluckiger asked if industry meant European industry and J Renuart's comment that it was the 'big 12' was felt to be restrictive. In ESPRIT the companies are from the 10 member states undertaking R & D but eventually companies in non member states could be looked at. P Linington saw the discussion getting confused as the rules of ESPRIT are not the same as members of the association. It was agreed that there was non ending discussions on what was European and for now it should be left for further discussion. K Ullmann did not see the problem. There is a clear goal to build OSI networks. All the 12 are doing it. It is not good if a way to cooperate between universities and industry cannot be figured out. B Carlson was confused on the membership issue and wanted a clear statement on who can join and who can take part.
P Linington thought there were two possible groups. Ether the EEC or COST signatories. The EEC excluded the Nordic countries and CERN and he thus considered COST a better grouping. M Hine thought the groupings irrelevant and thought it should be all academic networks committed to OSI. P Linington pointed out that this would admit the USA. M Hine was worried that the EEC would be worried if they saw EEC money being used outside of the community.
J Renuart said that the EEC did not want to restrict the scope for funding. C Zander opted for COST. If proposals are made for work then where it is done may depend on funding. The Scandinavians must be included as they are in COST. If they can get EEC money then so much the better. It was pointed out that Israel was not in COST but Switzerland and Austria were. M Hine said that there had to be interconnection with EARN they cannot be kept out of European academic networking. P Bryant thought that countries, such as Israel, could be given observer status which meant that they could have access to the activities and could take part but may not be able to accept funds and may have to pay in some way for collaborating.
P Linington pointed out that it may ne necessary to get things from outside the EEC and COST countries.
P Linington asked if the two decisions covered aims and membership. C Zander wanted a current work plan but not in detail.
M Hine thought the topics were:- Must go to ISO networks, Problems of transition and immediate interconnection. P Linington thought that some of the points in Zander's list were 10 years ahead and may not be OSI. The long term cannot be tied to ISO. C Zander replied that if one looked forward 10 years then one wanted to see standards develop fed by experience. J Renuart stated two points. In the short term to state an immediate intention to move to ISO, to define the actions needed such as gateways. The second point is to devise a migration plan together with CEN/CENELEC. M Hind thought all three should go together. K Neggers thought we must distinguish between the long and short term. The Association is long term. N Newman said that several ESPRIT projects were managed from within such as IES. J Renuart said that the academics must help themselves. P Linington said that things must progress rapidly. We must not get bogged down in detail. Proper use must be made of collaboration between network manufacturers and ourselves. We need funding if possible. C Zander said that cooperation was nothing new and it starts with working parties. It come from feelings that we must discuss. We must have a pragmatic approach. If we start with say 3 working parties it will give us some confidence we have something. B Carpenter thought that X400 was clearly an area for such work and the question was when not if. It is a question of resources. The critical resource is people. We have to get people to do the jobs. C Zander thought a lot of work could be done by mail. P Kirstein thought we should discuss this further tomorrow morning but Linington thought that the meeting of the organizing committee after the Workshop would be a better time. N Newman was worried about the resources. We need to find out what it will cost- including human resources. M Wells thought one solution was to persuade people to do it on behalf of the group what they were going to do anyway. There are a lot of things in common that we want to do. P Linington took it that the meeting wanted to progress things as fast as possible and this was agreed. The first priority was to organize the long term and this should be forgotten for this meeting. We must be certain that discussion do go on so it needs a working party. Other areas are basic interworking, mail and X400 are high on list- second party, LAN WAN interworking in another area. B Carpenter thought this to be very important but that the Association was to encourage interworking over long distances and that LAN WAN interworking did not encourage this. K Neggers pointed out that services could only be got to the users in many cases when the LAN WAN problem was solved. M Wells saw the connection of sites to national networks as a problem for the national level. That is not to say that the same mechanisms should not work in many places. P Kirstein saw much LAN WAN work going on in ESPRIT. He did not want to see the association diluted by this topic and it could easily get bogged down with it.
M Wells thought we had to get a relationship with CEPT. We do not want to get into the EARN problem.
P Linington asked if an X25 84 working party was needed and this was accepted.
B Butcher thought an information exchange system was needed. Could be termed Support of infrastructure group.
A Kundig thought we should not call ourselves network providers as this would provoke CEPT and the PTTs.
N Newman said that the commission had relations with CEPT via their telecommunications group and they could be involved.
A Patel thought that all the national networks should strive to have good relations with their PTTs.
F Fluckiger saw 2 or 3 areas for rapid work with deadlines for services. Should we also have a group with long term objectives. P Linington say a danger of multiple groups. The groups must be working groups and not committees. F Fluckiger also supported the mail and FTP initiatives in particular to exchange files between CERN and DFN.
A Kundig wanted means of exchanging electronic designs but P Linington thought that the activities should not become involved in research and design.
It was commented that the work should not duplicate the work of COST11 and there is some work which is best done in other groups.
P Bryant thought that the network managers should get together to exchange information on the operation of networks in the various countries.
N Newman wanted a directory of services.
P Linington was worried over addressing schemes. M Hine wanted to know what was happening about schemes for universal addressing and the reply from P linington was that there were several schemes.
B Mahon wanted directories of where things and people were but U Beyschlag was interested in mail directories.
P Kirstein thought that the addressing and mapping of addresses must be harmonized and that people interested in this could get in touch with the Thorn Collaboration.
P Linington suggested that we should use our best efforts to produce directory services and not get involved with research and P Kirstein agreed.
J Renuart wondered if there should be a support desk set up for the users. Linington thought there was a need to exchange information on topics in particular level 3 addressing, but B Butcher said this was covered in the X25 84 work.
J Prevost was interested in fostering work on satellite activities and E Valente could see that this was a mix between LAN and WAN technology. M Hine pointed out that there were several projects and it would be nice if the association could be a patron for these projects at least in word. He did not see a working party doing much for the long term. P Linington thought we should keep in touch with such activities but not put them on the high priority list and not duplicate the work of COST 11 and we do not want to do research. J Renuart supported this view and said that we were not replacing COST11 but we have to use the output of COST11. P Kirstein was worried over the broadening of the activities, a time come when experiment must end and we may want to put a satellite overlay on the current networks. However J Prevost did not think the satellite work was research and it was not only for HEP. J Renuart said that in ESPRIT they were not sure there was a need for satellite connections but can see a need for slow scan TV etc. We must concentrate on immediate needs. P Linington said that satellites were not high priority items. M Wells said he would get interested when the average traffic density gets above 5% on current links. P Linington suggested the item be put on the general interest list, it is not that there is no need but we have a lot of other activities but it will need investigation at some time. C Zander said that we had HEP groups who want high speeds. If there are other groups then there will be more pressure but we have enough to start with. J Renuart said that if we wanted momentum then we must get results quickly.
J Harms asked whether we should share the promotion activities. These could be coordinated internationally. P Kirstein thought otherwise in that this must depend on local circumstances.
F Fluckiger did not think any further work needed doing on the interactive protocols but directories would be needed. K Neggers wondered what services there were that could not wait for ISO protocols.
P Linington detailed the 8 priority areas.
1. Message handling and X400 2. X25 84 3. Collecting information on services 4. File transfer 5. Operation and management 6 Full screen services 7. Administration and forward planning 8. Liaison with CEPT
It was mentioned that information gathering was a major barrier to services. F Fluckiger thought MHS, FTP, Full screen, and information were the priority areas. J Hutton said that we must define urgency. B Carpenter said that we must find people to carry out the tasks. P Linington suggested that the 8 areas should put to the meeting the next day and asked how the next step should be organized. Are we asking countries to sign up and ask for participation? M Hine thought that it should be left to the network managers and B Carlson thought there should be a responsible person from each country. C Zander opened the question of whether there should be user groups but P Linington thought these would take a long time to find. B Butscher thought a committee should be set up tomorrow.
P Bryant asked whether we wanted EARN to be taken into account or not. J Renuart said that if we wanted to foster interconnection then EARN was not in but could be considered in the future. M Hine thought that if some sort of infrastructure was needed at an early date then EARN was needed. J Hutton suggested that they should be included via the national organizations. N Newman proposed that we did not need EARN formally. We could leave the formal issue of EARN open as regards membership of the association and for collaboration. B Carpenter said that we had to face facts that people think EARN is IBM run. If we want success we must not embrace EARN but we can cooperate with them.
B Carlson wanted to know how CERN and NORDNET could be handled. Linington thought NORDNET would be treated as EARN. We can solve these problems by only having nations as association members plus Euro organizations such as CERN. N Newman saw no problem with NORDNET.
C Zander raised the question of organization. He proposed that a management board be set up rather than we promote a program. Have a council formed from representatives of countries. B Carpenter thought an ad hoc organization group was needed. P Linington thought some arm twisting to get representatives was needed and to determine the next steps. This should go to the full meeting tomorrow to get ratified.
N Newman said that we must not go away without decisions. P Bryant thought that a permanent set of people should be found and financed to run the organization, some form of secretariat to ensure that things actually happened such as meetings and documents. N Newman suggested that the organizing committee should report in one month. We must start with a bang. P Linington agreed but thought 6 weeks more realistic.
J Renuart considered the calendar. The concept could be put to an ESPRIT meeting in mid July and it would be good if more details could be got by June. They need to be alerted that something was coming up. There was no time to loose. On funding- can only discuss finding when there is some detail on the organization and a work plan to which a budget can be put. Funding is likely to come mainly from the CEC. P Linington thought we needed some pump priming from COST and sub projects from the CEC. We should move towards self sufficiency. J Renuart saw several funding mechanisms. could be supported for 2 or 3 years and then self supporting but there must be an estimate of the budget. P Linington thought we needed to get the scope sorted out and then decide how to achieve it. N Newman thought the detail could take a long time to sort out and thus pump priming was needed. A reasonable budget was needed for June July. B Carpenter thought that the working groups should start off self financing. He mentioned that the policies of non member states may cause surprises. J Renuart thought this to be a naive question. He asked why network activities had not been funded before. P Linington noted that we had had meetings before and work was not being done. Now with a small increase in resources we can merge into a true Euro network. J Renuart said that if this was not the motive then money would do nothing.
It was resolved to present the meeting with the conclusions tomorrow and formulate actions.
Present:- A Patel Ireland M Walsh Ireland M Wells UK F Ros Spain K Neggers Netherlands B Mahon CEC J Harms Suisse A Kundig Suisse M Paul Austria M Hine CERN B Carpenter CERN F Fluckiger CERN E Valente Italy J Prevost France U Beyschlag CERN P Villemoes Denmark B Carlson Sweden P Kirstein UK B Butscher Germany J Hutton UK P Van Binst Belgium K Ullmann Germany N Newman CEC J Renuart CEC K Zander Germany D Barber UK P Linington Chairman P Bryant Secretary
The purpose of the European networkshop was to investigate the possibilities for providing good international networking services to the academic and research communities in Europe. The purpose of this meeting was to develop a proposal for continuing support of this aim to be put before the full meeting of the workshop.
There was unanimous support for the view that common and good networking services across Europe depended on the development and implementation of the ISO protocols. It was recognized that the successful implementation of these protocols depended on the harmonization of the options in the protocols selected as well as harmonization of items where decisions were needed such as naming and addressing.
It was agreed that any early activity should be directed to exploiting low speed (less than 64K bps) technology. Interest will be likely to shift to higher speed technologies as these develop and as lower speeds become well established and require less attention.
It was agreed that the primary responsibility was to provide networking services and it was not appropriate to support research and development which was not immediately directed to that end.
It was agreed that the networking services should be common to the academic community and to the industrial research community.
It was proposed to form an association. It was agreed that membership of the association should be on a 'country' basis rather than being an association of 'networks' or of 'interest groups'. It was agreed that membership, at least initially, should be confined mainly to Europe.
The association required to be connected with some existing organization in order to attract funds. The CEC was considered inadequate as it would have excluded the Nordic countries and this was unacceptable to the meeting. It was agreed that COST would be a suitable grouping of countries.
It was agreed that CERN should be a member in the light of its special international character, its strong requirements for networking and its networking expertise.
The COST grouping excluded countries such as Israel. It was agreed to create an observer status for such countries which would allow then to participate but possible not to receive funding or possible having to pay a subscription.
It was suggested that the association may need a small permanent secretariat to undertake organizational activities and secretarial services.
Professor Zander proposed the aims of the association to be:-
To form a powerful cooperative association To form a critical networking mass To harmonize networking To ensure convergence to standards To exchange experience To provide test beds with pilot projects To ensure feed back of experience to help develop standards To initiate new ISO standards where needed To stimulate the growth of new facilities such as super computer sites To stimulate and motivate action.
M Hine proposed three work items:-
The interconnection of the current ISO and non ISO networks to provide immediate services The migration of current networks to ISO standards the creation of ISO networking
It was considered important that services should be provided as soon as possible to create momentum and show results. The Migration to ISO protocols, which is also vital, will take longer to accomplish.
The proposed aims and work items were agreed.
It was unclear where finance would come from and how much was required. It was certain that early support would be needed for international working groups. It is to be hoped that the use of electronic mail and conferencing systems could reduce costs.
There was a suggestion that only modest finance would be required for development work as several national programs are already very active and the harmonization of the work was important rather than initiating new activities.
It was agreed to produce a financial plan in the near future and to seek suitable sources of funding. The CEC asked that some outline plan be developed as soon as possible for presentation to an ESPRIT meeting in mid June and that a detailed plan should follow.
It was agreed that the association should be directed by a group with academic representatives form each country. Prior to the establishment of the association it was suggested that the workshop organizing committee should draw up a proposal for immediate work items. It was unclear whether industry, PTTs, CEC or other groups should also direct the association.
It was agreed important that the object of the association was to foster networking and not the provision of a network. The association did not want to be seen in any way as a competitor to the PTTs. It was also vital to maintain good relations with CEPT as well as to foster good relations between national networking groups and their PTTs.
It was clear that the CEC wished to see industry taking advantage of the networking provided but it was unclear how the relationship between the two communities should start and develop. It was agreed that such relationships were important and to be welcomed. It was reported that after a few initial difficulties industrials and academics had worked well together in ESPRIT.
It was hoped that the association would be a powerful influence on manufacturers to provide the network components that were required. This must be achieved by close consultation and possibly by preferentially purchasing from companies providing products to the harmonized standards.
It was agreed that work should be started immediately and not wait for the creating of the association which could take a long time.
Eight priority areas were identified:-
Message handling and X400 X25 1984 Collecting information on services File transfer Operation and management Full screen services Administration and forward planning Liaison with CEPT
It was unclear what methods should be used in each area and it was agreed to ask the organizing committee to discuss this the next day.
It was resolved to put the above plan to the full meeting and seek their approval.
Present:- most of the organizing committee plus others who were interested
P Linington (Chairman)
P Bryant (Secretary)
The workshop had decided to set up an association to foster the provision of good networking facilities to the academic and research communities in Europe. They had also agreed to immediately undertake work in eight priority areas. This meeting has been asked to define what steps should now be taken to ensure that they are progressed.
It was agreed that in default of comment the current national representatives would stand as a provisional organizing committee of the proposed association. K Ullmann would now represent Germany. It was also agreed to take a relaxed attitude to the membership of the committee at this early stage, however for organization reasons there would only be one contact in each country who would be expected to liaison with any colleagues.
It was agreed that the activities of the committee would not be confidential.
It was agreed to ignore difficult political questions and leave these to a subgroup mandated to set up the association. That group would have to resolve the question of relationships with manufacturers, industrial research organizations, observer nations, ECMA, CEPT, CEC and COST.
The list of priority areas was considered and appropriate action taken in each case.
The amount of funding is unclear and N Newman agreed to attempt some estimates. N Newman
The provision details for the second European Networkshop are May 1986 in Copenhagen. B Carlson
The provisional details for the third European Networkshop are May 1987 in Spain. F Ros
At this early stage the various working groups will have to be autonomous and self supporting pending funding and an association. It is hoped that much of the activity can be conducted by electronic mail and the list of delegates includes mail addresses. If necessary it should be possible to set up conferences on KOM cutesy of QZ.
1 Organize X400 activity B Carpenter 2 Organize X25 84 activity Secretariat 3 Organize file transfer activity F Fluckiger 4 Organize full screen activity P Linington 5 Organize collection of information activity B Mahon 6 Organize network operation activity M Walsh 7 Organize drafting of association documents P Linington 8 Organize liaison with CEPT A Kundig 9 Produce funding estimates N Newman 10 Organize 1986 European networkshop B Carlson 11 Organize 1987 European networkshop F Ros
All European countries apart from Turkey, Yugoslavia and Portugal have public X25 networks. In the case of Portugal this only allows calls out of the country. In several countries the services may only cover restricted areas of the country.
In most countries there is some form of academic network services. These are based on a wide variety of technologies and a wide variation as to the proportion of the community served. Most countries have based existing services on X25 with a variety of high level protocols. Most of these networks use the triple X protocols for terminal traffic. A few had private X25 networks but all except the UK intended to use the public services in the future. Leaving aside EARN, many countries had small pockets of proprietary networks, in particular DECNET. This was particularly strong in High Energy Physics and in Italy.
Most countries had or were putting together national academic networking plans. Without exception these were based on the use of ISO protocols. Germany had a well advanced plan followed by the UK who had the problem of migrating an existing X25 network to using ISO protocols.
It had been difficult to get figures on usage, charging, availability and what percentage of the community could use the networks. It was clear that one of the principle uses of the networks was for mail and it is fortunate that this is an area of ISO standardization which is well advanced.
Besides the public X25 networks there are a number of other international networks. For example the banks and airlines have extensive ones and in the research field there are the Euronet and EARN networks.
The principle private physical network of interest to academics is EARN. This was set up in the Autumn of 1983 and is financially supported by IBM for 4 years. It uses the IBM RSCS protocols. There are about 100 nodes in Europe and 500 in its sister network BITNET in the USA. Several national networks, in particular SUNET, JANET and DFN intend to provide gateways to EARN. Although it only provides bulk services it is highly attractive as it is cheap and provides services to the states.
NORDUNET is about to start and is a network connecting the Nordic countries. It will be based on the existing telecommunications services and national networks. It will require the harmonization of standards within the region.
DIANE used to use EURONET but now is almost entirely on the public networks. It provides information services from 50 hosts which contain in all 600 to 700 data bases.
There are a number of international 'logical' networks which are groups of people who attempt to communicate over any networks they can. HEPNET was set up in 1981 by the High Energy Physics community and has been most successful in providing services with only modest developments. One of their principle activities has been the provision of gateways between various network technologies. These have usually been located at CERN. EUNET is somewhere between a physical and logical network. It is a network for mail between UNIX systems and tends to use dial up facilities. EUROMATH, like HEPNET, is a logical network for passing information between mathematicians which has just started.
ESPRIT promotes five aims. Advances in micro electronics, advances in information processing, software technology, office systems and computer integrated manufacturing. To help achieve this it requires a communications infrastructure called IES (Information Exchange System). IES has been heavily involved with standards. It is trying to provide mail, conferencing, text preparation, information retrieval, integrated software development and graphics for the benefit of the ESPRIT participants. ESPRIT is a 5 year program but may well be extended. The first call for proposals was in 1984 and a second call was in 1985.
The Rose project, under IES, is aimed at the provision of OSI protocols mainly on UNIX. The CARLOS project is aimed at the production of equipment to allow anything to be connected to anything. The THORN project will provide name server and directory facilities.
IES hopes to use existing network infrastructure and not to build a separate network.
RACE (Research in Advanced Communications Environment) is aimed at providing a 'broad band' communications environment by 1995. The first part of the project is to develop the ISO model for broad band. The second part is to do the research and development required to develop the equipment required. The topics for research are, high speed integrated circuits, high complexity integrated circuits, integrated opto electronics, broad band switches, passive optics, high bit rate components, long distance links, dedicated communications software and flat panel displays. The project will be subject to dynamic reassessment. It will be 50% CEC funded and cost 43M ECUs.
All PTTs were expected to migrate to X25 1984 but the time scale depended on the various plans of the PTTs, the plans of the manufacturers and also on the finance available. So far it is not clear what the interworking problems between X25 1980 and 1985 are.
Work was going on through CEPT to harmonize the actions of the PTTs. Hopefully the options implemented would be common to all European PTTs. Interworking problems were also being studied.
Three administrations were expected to have X25 85 by 1987 and the rest would follow at various dates.
In some cases features mandatory in X25 1985 were available already such as fast select. Other features such as 'transit delay select' and 'throughput class selection' needed some research to find out what they meant and how they should be provided. Some administrations will not provide some features but all will provide enough to support the ISO network service.
The use of the X29 parameters had caused a lot of confusion. It had been recognized that successful services needed some recommendations on the use of the parameters and several groups had worked on the subject. The UK had produced the Green Book recommendations which unfortunately used features only available on PSS. ECFA had done a similar study which stayed within the standard. This work had eventually led to the harmonization COS document. Green Book also defined a means for operating triple X over non X25 networks and this work had been extended with a draft proposal for putting triple X over ISO session layer. Meanwhile CCITT 1984 had revised the standard and further complicated the protocol. In CCITT 1977 there were 12 X29 parameters, in 1980 18 parameters and in 1984 22 parameters.
The ECFA group had found that a large number of triple X implementations were non standard or substandard in that they used parameters in illegal ways or failed to provide some mandatory parameters at all. The way the PADs and hosts negotiated the parameter settings often left a lot to be desired. The provision of national parameters was another area of difficulty.
The ECFA and COS work takes the pragmatic view that one has to recognize that many implementations are poor and so devise recommendations to make interconnection possible. Three modes of operation are recommended. The basic mode deals with equipment that only provides the mandatory facilities or often less than that. Advanced mode assumes that all the optional facilities are also available. The third mode is similar to the advanced mode and assumes that the 84 facilities are available. It was recommended that only equipment capable of advanced mode should be purchased.
Virtual Terminal Protocol (VPT) is not complete and an initial version is only going to be produced by leaving out many features where there have been difficulties. 6 classes have been defined, basic, forms, graphics, teletex, video and mixed. The basic class should be available soon as a draft standard.
There are still many issues to be resolved such as, 3270, other full screen services, performance of VTP, gateways between X29 and VTP, loosely coupled applications such as PCs and the future of quasi interactive protocols such as KERMIT.
A large number of file transfer protocols have been devised. In fact most networks have defined one. However ISO have almost finished the definition of FTAM which hopefully will be the last for some time.
There are, in fact 3 services required. The transfer of a single file. The transfer, examination and change of a file or file access. The creation and deletion of files or management. In addition services are not always between file store but may, for example, be to a job spool.
Most of the existing file transfer protocols only provided a subset of functions and often over a small range of lower level protocols. FTAM attempted to address all areas and, moreover use the lower level ISO protocols. There is some embarrassment that X400 uses different subsets of the session and presentation protocols. There are going to be problems in migrating from current protocols to the ISO one.
GIFT is one way of migrating. This project is aimed at providing a gateway between various file transfer systems. It currently or will soon deal with CERNET FTP, DECNET and Blue Book protocols. The project is realized on a VAX. It is hoped to incorporate UNINET at a later date. GIFT provides 'on the fly' file gatewaying. This was selected due to the difficulty of knowing whether a transfer had been successful if staging were used.
It should be possible to integrate FTAM into the product and so produce a valuable migration aid.
There are currently at least six mail systems in Europe- COM, UUCP,ARPA, CSNET and EARN. It would clearly be desirable to have one system and to base it on X400. Unfortunately few X400 systems exist and any service needs to be harmonized and uniform.
X400 contains two parts. The User Agent (UA) which prepares messages and accepts them and the Message Transfer Agent (MTA) which ensures that messages are sent to and received from remote MTAs. There is considerable flexibility as to where and how the UAs and MTAs are located.
It is possible to provide gateways between X400 and TELETEX and some administrations will do this. It is also possible to have gateways to TELEX.
A major problem is how the naming and addressing should be dealt with. X400 has naming domains and these have to be set up with some care for an organization such as the European academic community.
So far their are no commercial X400 implementations although a large number of manufacturers have developments under way. There is a version of X400 called EAN which has been produced at the University of British Columbia and is on trial use at several sites in Europe.
In one year it should be possible to have an X400 node in each country. It would be necessary to have a directory service in each country. To do this some form of European cooperation would be needed to ensure that the system would provide services across Europe.
The rest of the world is principally what is happening in the USA where there are some 130 networks. Mail is one of the main networking activities and many gateways exist between the networks to provide services across networks.
USENET is dedicated to mail and has 2000 nodes worldwide. It uses the public telephone network and is run on a cooperative basis by the UNIX community. CSNET and MAILNET are also dominated by mail and have some hundreds of subscribers.
In most cases gateways work but they often provide poor failure notification facilities and directory services are poor. In fact no gateways support any nameserver functions yet. There is little name registration to help users. In fact it is rare to find a good gateway as they are usually grafted on to machines providing other services.
Most of the networks do not allow transit traffic which is often needed because of the ad hoc topology of the networks.
An attempt is being made to set up 'Science NET' which is aims at connecting federally funded resources, in particular super computers.
Commercial networks are developing fast and are likely to provide X400 services soon. Some of these have international services and others are forming partnerships with other suppliers.
ISO in addendum to 7498, ECMA in 'OSI Directory Access Service and Protocol' and CCITT in study group VII have recognized the problems of naming addressing and directory services.
Since directories will have to be distributed the access to directories will have to be via networks and thus require a suitable protocol. The user will first access a local 'Directory Service Agent' (DSA) using a 'Directory Access Protocol'. If the DSA cannot satisfy the request then it will enquire of other DSAs using a 'Directory Server Protocol'. Work is going on to standardize the information which appears in directories. Several types of service are required such as Information services, white page services and yellow page services. Updating services are also needed.
To make such a scheme work there will have to be naming authorities to ensure that names are unique and to provide conventions for naming and to manage it. To achieve this there will have to be a hierarchy of naming domains.
There are still problems to be solved with authentication and consistency. The Esprit Thorn project, and some projects within DFN are attempting to progress the provision of directory services.
The performance and failure reporting mechanisms in both private and public networks are poor. The PTTs complain that they suffer from lack of information from their equipment on national networks and the situation is worse with international links. In fact 5% to 7% of calls fail and this figure is rising. It appears that there are only primitive means for testing links and no dynamic reallocation on failure. The situation is made worse by the error messages on various networks being inconsistent.
It appears that there is little active monitoring of the public networks ether by the PTTs or customers. There is clearly a need to improve this situation and the academic community would do well to closely monitor the performance of the public networks and to have very close liaison with the PTTs.
The tariffs of all European PTTs are in the Eurodata Foundation Yearbook which, although expensive, is a mine of information. In most cases it is the charge per kilosegment which dominates. The charges across Europe vary considerable for the various services in each country. However there was no 'overall winner' as a country that had a cheap volume tariff might well have a high duration charge.
There are many questions that the community could ask the PTTs. Is the international surcharge reasonable? Why are there no discounts for night and week end use? Should there be bulk tariffs? Should the academic community have a special discount perhaps on account of possible collaboration with the PTTs?
An option that may be available is to base a European academic network on leased lines which should control costs and hopefully give a cheaper service. However this would require manpower, would incur large expenditure now and again, there would be legal problems and it may be difficult for small countries to participate. One could also ask why it had not been found necessary to have an academic telephone network?
The means of charging varied from institute to institute. If costs were small then charging could be part of general computing but it is possible for a user, possibly by accident, to incur high charges.
The principle problem seemed to be how to persuade the PTTs to charge a tariff that academics could afford.
ISO protocols start as working papers. They then become 'Draft Proposals' (DP). This is followed by 'Draft International Standard' (DIS) and they finally becomes an 'International Standard' (IS). Standards are usually quite stable by the time they reach DIS stage.
The current dates for ISO protocols are:-
FTAM end 85 for DIS JTP now in second DP Presentation now in second DP Session is an IS Transport is an IS Network is a DP.
The CEC became interested in standards in 1977. In 1979 the 'multi annual program' funded COST and CREST to further standards and undertake research and development for public purposes. In 1984 there was a Senior Officials Group on IT standards. In 1985 CEN/CENELEC became interested in the use and harmonization of standards.
In 1983/4 there had been an initiative to harmonize standards for academic use which had resulted in the COS documents. This work was taken up by SPAG and produced the GUS documents (Guide to the Use of Standards).
In 1984/5 there was CEN/CENELEC - CEPT harmonization.
CEN/CENELEC mechanism is that the CEC mandates a piece of work. After the technical work the document goes for public comment and then for ratification. It then becomes a Euro norm.
There are 5 priority areas. OSI level 1-3 and triple X. OSI level 4-7. Document transfer and messaging. Formal descriptive techniques. Programing languages, GKS and COBOL etc.
In 1985 there was a draft commitment directive on IT standardization following a directive on procurement. The work follows the following pattern. Determine the basis for standardization. Add precision. Develop new standard. Determine application of standard. Demonstrate developing norms. Promote the use of the IT standards. Management committee.
In fact 90% of the standards activities is in networking.
Currently JANET operates on the Coloured Book protocols. It is intended to migrate to ISO on an application by application basis rather than on a layer by layer basis. This will giver fewer transitions. Various relays will be required to communicate between machines running old and new protocols.
In migrating to ISO network layer the network address will have to be rebuilt and it is intended to use the ISO DCC scheme and JANET is applying to the 'competent body' for numbers.
It is intended to apply pressure to manufacturers to provide all the network products rather than producing then in the community.
Currently all machines for general use in the community must provide Coloured Book protocols. This procurement policy has paid dividends in promoting JANET. Soon the procurement rules will allow suppliers to provide ISO products instead of Coloured Book ones and at an even later date ISO products will be mandatory.
The complete transition of JANET will take between 10 and 15 years.
There are now 70 machines in DFN which is based on the public X25 network. The services provided are triple X, file transfer and job transfer. The machines are VAX VMS, Siemens/BS2000, PDP11/RSX, ND100 and Siemens/BS3000.
DFN intends to use EAN. The migration to FTAM would take place in 1987 and to JTP in 1988. It is hopped to use manufacturers products wherever possible. One of the problems will be guaranteeing interworking. Another problem is the provision of stable user interfaces.
The European Academic Research Network (EARN) was set up in 1983/4. It uses the IBM RSCS protocols and operated over leased lines. The protocols limit it to file transfer and mail activities. There are about 100 machines connected and a further 500 machines on its sister network BITNET in the USE. The two networks are connected via two leased lines.
EARN is set up as an association set up under French law. It is run by a Board Of Directors, one from each country, who meet a few times each year. The international lines are paid for by IBM for 4 years but the network is controlled by the Board of Directors and IBM only contributes finance and advice.
Since EARN indulges in third party switching it requires licenses or permission from the PTTs to operate. In fact CEPT has taken a special interest in the network and has recommended that EARN must migrate to use ISO protocols and the public networks within some reasonable time scale. CEPT also recommended that the PTTs impose a volume tariff on traffic. In fact the tariffs imposed vary considerably with many countries imposing a zero volume charge and merely the leased line charge to the other extreme of imposing a tariff which is close as possible to the IPSS charges together with a fixed charge which approaches the leased line charge.
EARN has a technical group which meets from time to time to direct the technical development of the network. A small subgroup is studying how the network should migrate to ISO protocols. To this end they will soon be having discussions with IBM technical experts to see how best the migration should take place. There turns out to be several options from academic developments from Darmstadt and Salford University to ones based on IBM developments.
In Italy there is a policy of putting in gateways between the various networks. These are- X25, DECNET and EARN.
It is intended to put in two leased backbone connections. First- Rome, Pisa, Madrid and Florence. Second- Rome, Florence and CERN. The public networks would only be used for low volume traffic.
ISO standards would be used as they became available with performances comparable to the currently used ones.
48K lines are expected in a year with 128K via satellites available later.
17 Nordic Transition. L Backstrom.
NORDUNET starts this year and is a transition program based on the national networks.. It is expected to take 4 years to produce a coordinated infrastructure providing a reliable service. There will be a harmonization of standard strategies and new value added services for end users. it has funding of 10M SEK.
The networks involved are:-
Denmark with CENTERNET on private X25 lines and 3 hosts and terminal traffic. Finland with FUNET on public X25 lines and 11 hosts and terminal traffic. There is also a DECNET network. Sweden with SUNET on public X25 lines and 17 hosts and terminal traffic. Norway with UNINETT on public X25 lines with local switches and 22 hosts. Traffic is terminal, file transfer and MHS. Iceland has yet to start.
COST 11 ter follows COST 11 bis which followed COST 11. Ter like bis lacks the focus of a support network. Derek Barber was Director of COST 11 and Thomas Kalin director of COST 11 bis. The project leader for COST 11 ter has yet to be announced, although a recommendation has been made to the Commission by the Consultation Committee. The committee is responsible for managing COST 11. There is also a higher committee responsible for 'Manufacturer application Program' from which the budget is taken. The ter budget is small at 2M ECU plus small contribution from non Community countries for R and D. 1.4M ECU is allocated to COST 11 and .4M ECU is so far un allocated. Several of the projects are carried over from bis.
The countries involved are:- Belgium, Denmark, Spain, France, Switzerland, Italy, Great Britain, Ireland, Norway, Sweden, Germany and Yugoslavia.
The projects are:-
Open shop information services. Bengt Olsen, Sweden. Human computing services which follows on from GILT and is interested in X400 protocols. Rolf Speth, Germany. Formal description techniques. Aldo Le Mole, Italy. Distributed systems management. Alwyn Langsford, UK. Human factors. Willy Jensen, Norway. Security mechanisms for computer networks. Not yet accepted.
COST needs strengthening. ESPRIT, RACE and ALVEY are exploitation and industry orientated and they are using the results of earlier research which has sometimes come from COST. This research activity is at a low level and the pendulum must swing back to provide more research.
There are three topics considered, the user needs, the advances in technology to meet the needs and the PTTs' plans. The users are the research community and those who wish to communicate with them over long distances. The ideas for future communications come from various conferences held recently including the ICCC 'Links for the Future' in 1984 and the COST 11 seminar in Rome.
The current user facilities are all limited by speed and better links will allow the following facilities to be offered- full screen graphics, large file transfers, fact fax, mixed text and graphics, slow and normal scan TV, audio, conferencing, broadcast, mobile and temporary working.
Current facilities, such as COM, are a good step forward as are X400. It is already cheaper to send an A4 page via X25 than by post but for 10 pages it is not true. On broadcast facilities the PTTs offer no useful services. Megabit per second working is attractive and ought to be important for everything as long as the tariffs are reasonable and the regulations for use reasonable. The universality of services is important for our community.
Technical improvements will offer 50K to 2M bps in next 10 years which will satisfy a large fraction of the academic users however one must remember that a single LandSat picture contains 1 Gbit of information. There are 4 media, copper, radio, satellite and fibre optic, which will all still be in use in the year 2000. Currently the use of technology is 70% analog, 10% digital coaxial, 10% digital radio and 10% fibre optic. By 1988 this will have changed to 50% fibre optic, 30% coaxial, and 20% digital radio. Most data currently goes over analog links which often go over 64K PCM links multiplexed over 2, 8.4, 34, 139, and soon 565 Mbps channels. A new CCITT standards allows voice circuits over 32K channels. The PTTs see most of their revenue coming from voice in the future.
Although fibre optics are taking over many applications others such as TV will remain analog since a digital television picture needs 80Mbps whilst an analog one only takes 6Mbps.
Several new possibilities will become available with satellites. It will be possible to transmit from the satellite a narrow beam 100 miles wide which may allow some forms of switching. Another interesting possibility is to launch light aircraft which fly at high altitude and form a radio platform. These will act almost like a satellite but without the delay.
Fibre optics can now stretch for 30-40 Km without repeaters at 140Mbps. The next step should be 85 Km between repeaters at speeds of 565Mbps. In the laboratories distances of 200 Km at 0.5Gbps are possible. An interesting technique is to put several lasers on a single fibre thus giving a broad band capability.
The PTTs are moving towards digital services and ISDN. BT expect to be fully digital by 1990 but subscriber links will not be digital until 2000 but they will be available for those who need them. It is difficult to see what the impact of ISDN will be and in fact the PTTs themselves do not know.
A basic problem is that the PTTs primary role is to provide speech services and digital ones form a small part of their business and it is grafted onto the speech network. Thus the measures of acceptability as regards error rates are not meaningful for digital applications
The PTTs still think in national terms and it should be a task of this workshop to persuade them otherwise.
The users want to do their normal work without spending any significant amount of time learning how to use networks. Since they are not and do not want to be network experts they are unclear how the networks will contribute to their work in the next 2, 5, or 10 years from now. It is therefore difficult to get an estimate of the users medium term needs. They always want money for their research needs and are thus reluctant to spend much on networking and thus need cheap services.
As well as new services the users need better versions of existing services.
The interactive services should provide more performance, in particular good full screen services together with editors are needed. There could be advantage in providing 'standard' methods of use on machines. There needs to be a solution to the problem of where to edit file- are they edited on the remote machine or file transferred to a local one first. Interactive graphics are needed which require a good bandwidth.
Scientists would now like to control their experiments remotely to avoid wasting time and money in travel.
Electronic mail must be easier to use and generally available. The users should not need to know about the details of the networks but only the name of the recipient and possibly the site he is at. Mail should be able to deal with formatted text, graphics and other information so that it can be used as a simple form of file transfer which does not require a knowledge of the remote file store. Although EAN is a step in the right direction, implementation of X400 on a wide range of machines as commercial products are needed. The need for mail across Europe is urgent.
There seems little need for digitized voice.
Computer conferencing is widely used and popular in some communities. It does seem that it suffers from the problem that users have to access a remote machine to use it and would rather conference entries could come to their home machine. These systems also suffer from not providing a wide character set.
Tele or video conferencing is not used much and there seems little need for it. Its disadvantage is that the participants all have to be present at the same time and for a limited period. This technology does not seem to encourage 'coffee' style discussions which are important. The problem may be in the limitations of the equipment rather than the basic facility.
File transfer falls into two groups. First there are small file such as programs, data samples and graphics of less than 5M bytes. Second there are the large files such as experimental data and digitized pictures of 20 to 200 M bytes. In both cases the users want reliable, cheap and responsive services although large transfers may well be less urgent and can be transferred at off peak times when tariffs are low. As soon as it is possible to transfer large files the service will be heavily used as long as tariffs are reasonable.
Services to transfer text which is capable of re editing and re formatting are needed as a lot of traffic will be directed towards the production of documents. As yet there are no standards apart from the de facto IBM Script one.
Users seem happy with the current remote batch processing facilities based on the 'Red Book' or IBM protocols. Theses services are heavily used.
There are problems with the concept of the scientific workstation. Currently these are small machines with alphanumeric or graphics screens and slow links into remote machines. In the future the computing load needs to be shared between the work station and the central machine so that the tasks are done on the machines best suited to the jobs. A medium speed reliable network is needed for this service but it is unclear what it should be. Can the digital telephone services be used for example. It may be appropriate to try and distribute must of the computing load to machines located in various places on a site.
Scientists often like to work out of hours or from home. The principle problem is to have fast links to the central site and the current 300 or 1200 BPS services are too slow and unreliable.
The 1200 to 64000 bps traffic can be expected to increase in particular on international links within Europe. Higher speeds of 512 to 2000 K bps are needed now and equipment to deal with these speeds is also needed. To deal with such speeds high speed digital terrestrial links, satellite services and X21 connections are needed. The PTT services should be cheaper, less rigid on the services provided.
The applications provided should be to ISO standards and these standards are needed quickly. The interpretation of the standards should be done carefully to ensure interworking across Europe is possible. To achieve this effective European coordination is needed. Pressure is needed on manufacturers to provide these products. In one or two years time products should not be bought from suppliers not providing suitable products. Gateways are needed to communicate with networks which use less standard products.
The main restriction on traffic is the high tariffs and this conflicts with the need to put more money into the science itself. A special tariff for academic traffic would be very welcome. In particular the removal of the premium on international European traffic would be helpful.
Users needs are growing and the principle requirements are for mail, file transfer facilities on high bandwidth, reliable networks which follow standards. The networks must be coordinated across Europe. The tariffs of the PTTs are currently too high and need to be reduced if the user needs are to be met.
The closing session of the European Networkshop was led by Professor Zander and discussed the future activities required to provide a networking infrastructure for the academic and research community.
This week had seen the bringing together of colleagues from the various parts of Europe. It had been an interesting meeting and we must now decide how to proceed, how to harmonize the activities in order to converge our aims. We have had discussions and now we must recommend. The meeting has shown that there is a lot of common ground and we should be able to do something useful.
We should establish an association. We should create an academic infrastructure. We should not be providing networks which is the task of the PTTs. We must have some aims and some scope defined. There must be some short term and some long term objectives to provide high quality networking.
The scope must be Europe, EEC and COST. It must include the whole community including the academic institutions and research laboratories. The activity must maintain and gather speed. The European industrial research laboratories could be added later and the user groups should also be associated with the activity. ESPRIT is an important group of users who can benefit and the facilities must be available for academic and industrial collaboration.
A program is needed which must start now and not wait for the establishment of an association or for funding. The critical items must be identified and problems solved in parallel with setting up more permanent support. Some support should be available from national programs and one can also look to the CEC and COST for support. The association must be set up on a sound footing under the laws of some country.
The trend of the activities should be to create international OSI networking, to migrate existing networking to ISO standards and also to foster the short term interconnection of existing networks.
The priority areas which were discussed at yesterdays meeting are:-
X400 messaging X25 1984 File transfer Full screen support The collection and dissemination of information Exchange of operational information The development of the association Liaison with CEPT.
We should have a further Networkshop in mid 1986 to give some focus for the working groups when progress could be assessed.
We must now decide is we want an association. We must see who can be expected to participate. We must decide on priorities. We must identify the people who will progress the objectives.
The meeting, in a subsequent discussion were unanimous in supporting the initiative.
It was thought essential that the PTTs and CEPT must be made fully aware of the activities and must be given no grounds for thinking that the work would in any way encroach on the responsibilities of the PTTs but should rather cooperate with them and exploit the facilities they provide.
It was agreed that membership must not be restricted to members states of the CEC but must be widened to COST and means of cooperating with other countries must be found. However, and arrangements had to be agreeable to the funding and sponsoring bodies. It was also agreed that funding was principally for fostering collaboration and not for development projects and thus modest sums for travel and a secretariat were needed.
M. Reuart welcomed the initiative and stated that ESPRIT and IES were anxious to see the association set up particularly as the academic community had a strong roll to play in ESPRIT. They wanted to make use of the infrastructure being pioneered.
It was important for the organization to gain the confidence of everyone inside and outside the community. The work must be open minded for open networking. There must be no compulsion to join the association but those who did would be demonstrating their willingness to support and contribute to the associations' aims.
The meeting unanimously endorsed the setting up of the association and the initiation of work in the priority areas.
A workshop of representatives from the European academic networking communities was held in Luxembourg from 13 to 15 May 1985. The meeting, was sponsored by:-
COST - Cooperation Europeenne dans la Domaine de la Recherche Scientifique et Technique.
ECFA - European Committee for Future Accelerators (which co- ordinates High Energy Physics research in Europe).
ESF - European Science Foundation.
The meeting resolved to set up a European association to coordinate and foster the development of networking within Europe for the benefit of the academic and research community. This has subsequently been provisionally named RARE - Reseaux Academiques et de Recherche Europeens.
Networking facilities have been introduced into most academic and research institutes in Europe. A variety of technologies have been used depending of the manufacturers involved and the particular circumstances at any site or within any country. This has led to difficulties in communicating between computers in different countries and often within countries. Recognizing these problems, many countries have initiated national programs to harmonize the activities within their borders.
Professor Karl Zander, from the Hahn Meitner Institute in Berlin, recognized that in setting up the German DFN project there were requirements for communications traffic between countries and in particular between the European ones. He therefore initiated a series of meetings in late 1983 and early 1984 to attempt to harmonize the protocols planned to be used in the various countries. This work was generously supported by the Information
Technologies and Telecommunications Task Force of DG3 of the European Commission. This work was based on the emerging protocol standards being developed under the auspices of the International Standards Organization and by CCITT. It was realized that good communications facilities between the large variety of equipment already installed could only be achieved by the use of a set of non proprietary, standard, and adequate protocol implementations. Such a set could only be derived via ISO. The result of this activity was a set of documents which defined inter working subsets and implementation recommendations for many of the ISO standards or draft standards. These documents have been used extensively in some ESPRIT activities and within several of the national academic network groups.
In late 1984 advice was sought as to whether it was appropriate to hold a 'workshop' to consider further harmonization and collaboration. The idea was enthusiastically welcomed and an ad hoc organizing committee started to plan the venture. Sponsorship from suitable 'academic' organizations was obtained. The organizers met and decided that the workshop should be limited to 60 or 70 people to ensure good discussion. Attendance was limited to delegates representing countries and representing the sponsors. To this end, a representative was selected in each country who selected the rest of the national delegation. The size of a delegation reflected the population of the country.
The European Commission generously hosted the workshop in Luxembourg and provided organization resources. The location also reflected the pan European nature of the workshop.
The program reflected the desire to concentrate on the exploitation of existing and emerging technologies for the benefit of users rather than to consider research topics.
The workshop was opened by Professor Karl Zander who stressed the importance of achieving a good network infrastructure for the European community. He outlined how he thought this could be achieved. He said that this was a very appropriate time to be starting an initiative and the goals were within reach. His address is reproduced in annex 4.
During the evening of May 14 an informal meeting was held involving country representatives and the sponsors. The meeting proposed that an 'association' should be set up to foster European academic networking. There were strong indications that funds could be found from one or more of the European institutions to support such a venture. In the main these funds would be needed for travel, meetings, and perhaps a small permanent secretariat. The main purposes would be for cooperation, collaboration, and harmonization. Pending the setting up of an association it was proposed to set up initiatives in eight priority areas namely:-
The minutes of this meeting are in annex 6.
The proposals were put to the workshop in the concluding session and met with unanimous and enthusiastic acceptance. A record of the concluding session is in annex 3.
After the workshop the organizing committee met and considered how the eight priority areas should be progressed. A record of their discussion is in annex 2.
A number of presentations were given during the workshop which defined the state of networking and explored a large number of options for the community. The details of these will appear in proceedings. A summary of each talk is given in annex 5.
An address list of the participants is given in annex 7.
The annexes attached are:-
1. Proposal for the establishment of a European Association to promote the creation of a unified network infrastructure for the support of research and academic collaboration.
2. Minutes of a meeting of the organizing committee of the European Networkshop held 15 May.
3. European Networkshop - closing session.
4. Opening address to the European Networkshop. Professor K Zander.
5. Summary of presentations given at the European Networkshop.
6. Minutes of an informal meeting involving country representatives and the sponsors. 14 May 1985.
7. List of participants.
Further copies of these documents may be obtained from the address on page 1.
National academic networks now exist or are being planned in a number of European countries. Almost all of these networks are either based on, or give access to, the PTT provided public data networks, so that communication between these networks is possible. There is a growing interest in the use of these networks to support international collaborations.
The major barriers to the creation of an effective European infrastructure are the lack of information and the lack of harmonization of means of access to the available services. To try to overcome these lacks, a European Networkshop was set up on an ad hoc basis by a group of those responsible for providing and using the existing networks. Support for the activity was sought from COST 11, the European Science Foundation and the European Committee for Future Accelerators. This workshop was held in Luxembourg from the 13th to the 15th May 1985. The aim was to compliment the previous initiatives taken by Professor Zander in protocol standardization, and promote the practical use of a network infrastructure. About 70 people attended; 17 countries, the CEC and CERN were represented.
The workshop covered a large number of areas of concern in the establishment of a unified infrastructure, from review of the current situation to the discussion of standardization requirements, but the main conclusion was that further and continuing action would be required if the necessary harmonization was to be achieved.
In the concluding session of the workshop, it was agreed unanimously that an association should be established to promote the creation of a European research and academic networking infrastructure. The principles of Open Systems Interconnection should be applied to ensure the widest applicability.
The aims of the activity will be to provide a high quality networking infrastructure for the support of research and academic endeavour on a European basis, using the public data services provided by the European PTTs in a harmonized way. The association will take any necessary actions to ensure that this infrastructure adopts and exploits the most advanced technology available.
The infrastructure should cover the whole of Western Europe (which was taken to be the EEC and COST11 countries). Membership of the association is expected to grow; the initial nucleus has been formed from the national academic networks and the major European Research Laboratories, represented by CERN, since these currently have the most advanced common infrastructure, and will benefit immediately from its enhancement. To this nucleus can be added a widening circle of European Industrial Research Centres since these share with the academic institutions a common need for tools to support collaboration. Indeed, it is hoped that this joint approach will complement the ESPRIT objectives of strengthening collaboration between industry and the academic world. It is also hoped that representatives of specific subject oriented user groupings (such as Mathematicians, Biotechnologists, Computer Scientists, etc) will become involved.
Several distinct threads of activity have been identified, corresponding to different timescales. These are:
The relation of ESPRIT to the activity was discussed at some length and deserves special mention. It was agreed that the ESPRIT community was a particular grouping of network users which would benefit greatly from the rapid creation of a common infrastructure. As such, it was an outstanding example of the type of community this initiative was intended to assist, and could be looked to for assistance in achieving the common aims.
Certain items were identified by the workshop as requiring immediate action. These were selected on the pragmatic basis of short term need and available resources. For each item, a lead organization or country was identified and charged with ensuring that the progress is maintained. The initial items are:
The association: The creation and progression of proposals for the support, constitution and the longer term organization of the Association. This will include liaison with COST11 and the CEC. (The UK will lead this activity.)
Liaison with CEPT: This activity covers the scope and mechanism for liaison between the association and CEPT. The aim is to establish a significant user voice in the discussions of the public communication infrastructure for Europe. (Switzerland is to be asked to lead the activity.)
Exchange of Information: It was agreed that a regular forum was required to allow for exchange of information between the association members. This would be achieved by establishing the European Networkshops as a regular annual event. Sweden volunteered to host the event in May 1986, and Spain the event in 1987.
Message handling systems: This activity will be based on the CCITT X.400 series of recommendations, and is to cover harmonization of options and facilities to be provided by message handling systems. In the short term the group will coordinate the establishment of a pilot community based on the EAN implementation. (CERN to call an initial meeting, one function of which will be to determine the longer term responsibility. Sweden and FRG expressed willingness to take a major role).
X.25 (1984): This activity is to examine the user requirements and timescales for the provision of the features of X.25 (1984 version) which have been added to support th4e OSI Network Service. The aim is to make input to CEPT on the user view of requirements. (France to lead the activity).
File Transfer: This activity covers both the short term interworking between file transfer systems and the adoption of the OSI File Transfer, Access and Management standards. (CERN to lead the activity).
Full screen, terminal working: There is currently a lack of suitable standards for the support of screen oriented activities, such as editing, across the public packet networks. The aim of the activity is to identify a suitable open solution. (UK to lead the activity).
Collection of Information: The aim is to collate information from the various national networks, and provide initial European directories. This process will be manual; the result will not be complete, but will represent an enormous advance over the current vacuum. It will cover
(CEC will lead the activity).
Exchange of operational information. The aim of this activity is the transfer of operational experience and information between the current academic networks. The scope of the activity is limited to activities up to and including layer 3 of the OSI reference model. (Eire will lead the activity).
Present:- most of the organizing committee plus others who were interested
P Linington (Chairman) P Bryant (Secretary)
The workshop had decided to set up an association to foster the provision of good networking facilities to the academic and research communities in Europe. They had also decided to immediately undertake work in eight priority areas. This meeting has been asked to define what steps should now be taken to ensure that the objectives are progressed.
It was agreed that in default of comment the current national representatives would stand as a provisional organizing committee of the proposed association. For organizational reasons there would only be one contact in each country who would be expected to liaise with any colleagues, at this early stage documents would be sent to more than one person in each country if necessary.
It was agreed that the activities of the committee would not be confidential.
It was agreed to ignore difficult political questions and leave these to a subgroup mandated to set up the association. That group would have to resolve the question of relationships with manufacturers, industrial research organizations, observer nations, ECMA, CEPT, CEC and COST.
The list of priority areas was considered and appropriate action taken in each case.
The amount of funding is unclear and N Newman agreed to attempt some estimates. N Newman
The provision details for the second European Networkshop are May 1986 in Copenhagen. This will be in conjunction with a Nordic academic networking conference. B Carlson
The provisional details for the third European Networkshop are May 1987 in Spain. F Ros
At this early stage the various working groups will have to be autonomous and self supporting pending funding and an association. It is hoped that much of the activity can be conducted by electronic mail and the list of delegates includes mail addresses is appended. If necessary it should be possible to set up conferences on KOM courtesy of QZ.
1 Organize X400 activity B Carpenter 2 Organize X25 84 activity J Prevost 3 Organize file transfer activity F Fluckiger 4 Organize full screen activity P Linington 5 Organize collection of information activity B Mahon 6 Organize network operation activity M Walsh 7 Organize drafting of association documents P Linington 8 Organize liaison with CEPT A Kundig 9 Produce funding estimates N Newman 10 Organize 1986 European networkshop B Carlson 11 Organize 1987 European networkshop F Ros
The closing session of the European Networkshop was led by Professor Zander and discussed the future activities required to provide a networking infrastructure for the academic and research community.
Professor Zander said that this week had seen the bringing together of colleagues from the various parts of Europe. It had been an important and interesting meeting. We must now decide how to proceed and how to harmonize the activities in order to converge our aims. We have had discussions and now we must recommend. The meeting has shown that there is a lot of common ground and we should be able to do something useful.
We should establish an association. We should create an academic infrastructure. We should not be providing networks which is the task of the PTTs. We must have some aims and some scope defined. There must be some short term and some long term objectives to provide high quality networking.
The scope must be Europe, CEC and COST. It must include the whole community including the academic institutions and research laboratories. The activity must maintain and gather speed. The European industrial research laboratories could be added later and the user groups should also be associated with the activity. ESPRIT is an important group of users who can benefit and the facilities must be available for academic and industrial collaboration.
A program is needed which must start now and not wait for the establishment of an association or for funding. The critical items must be identified and problems solved in parallel with setting up more permanent support. Some support should be available from national programs and one can also look to the CEC and COST for support. The association must be set up on a sound footing under the laws of some country.
The trend of the activities should be to create international OSI networking, to migrate existing networking to ISO standards and also to foster the short term aim of interconnecting existing networks.
The priority areas which were discussed at yesterdays meeting are:-
We should have a further networkshop in mid 1986 to give some focus for the working groups when progress could be assessed.
We must now decide if we want an association. We must see who can be expected to participate. We must decide on priorities. We must identify the people who will progress the objectives.
The meeting, in the subsequent discussion were unanimous in supporting the initiative.
It was thought essential that the PTTs and CEPT must be made fully aware of the activities and must be given no grounds for thinking that the work would in any way encroach on the responsibilities of the PTTs but should rather cooperate with them and exploit the facilities they provide.
It was agreed that membership must not be restricted to member states of the CEC but must be widened to COST and means of cooperating with other countries must be found; however arrangements had to be agreeable to the funding and sponsoring bodies. It was also agreed that funding was principally for fostering collaboration and not for development projects and thus modest sums for travel and a secretariat were needed.
M. Renuart welcomed the initiative and stated that ESPRIT and IES were anxious to see the association set up particularly as the academic community had a strong roll to play in ESPRIT. They wanted to make use of the infrastructure being pioneered.
It was important for the organization to gain the confidence of everyone inside and outside the community. The work must be open minded for open networking. There must be no compulsion to join the association but those who did would be demonstrating their willingness to support and contribute to the associations' aims.
The meeting unanimously endorsed the setting up of the association and the initiation of work in the priority areas.
In 1981/2 when we started to build the DFN network in Germany, which is a network for research, we soon realized that we needed harmonization between the networking programs in Europe in order not to make the mistake of having incompatible networking across the continent. We saw that different national and European programs had been started to develop open information exchange systems for wide user communities. We wanted to reduce the intellectual and financial effort engaged in these developments and promote the development of national compatible networking in the different member states of the European Community. This motivated the requirement for harmonization between the different programs to provide a trans national information exchange system in Europe.
The purpose of the European harmonization action initiated by me in October 1983 was to harmonize the use of information technology standards commencing with those for open systems interconnection. This harmonization was needed to ensure thorough compatibility between networks that were being developed in different countries on a national as well as European basis. This would encourage the European manufacturers to develop products to be used in conjunction with all or any of these networks, to eliminate parallel developments and to eliminate the waste of financial and intellectual efforts.
My feeling is that it is necessary to make common recommendations on the selections of options and classes from the ISO standards and their modes of use. In many cases the standards are not available in full international form and it may sometimes be necessary to base these harmonization upon ones which are only at draft international standard or draft proposal stages of development. This recognizes that some suppliers will begin to develop products in anticipation of the complete standards and, therefore, to have the greatest benefit harmonization of these developments should begin as early as possible.
Therefore, the major purpose of this European activity is harmonization of the community communications market and the stimulation of the community industry to supply this market. The harmonization must be such that the resulting recommendation can be quoted in CEC procurement specifications.
I think this workshop supports this idea in meeting to discuss the several routes for implementing the network standards.
A key role, in my thinking, is that we should think of the interconnectability of Europe wide networks, which avoid translating gateways, as a common symbol of interconnectability. One which over rides barriers not only in networks, multi national networks, and their multi vendor equipment, but establish a tool for cooperation of human individuals. This is exactly what Europe needs ladies and gentlemen.
The key words are- communication, cooperation, and concurrence on the basis of common information standards and recommendations.
This has given a short introduction to what we called, at that time, 'The European Harmonization Activity' or EHA. The participants came from Denmark, France, Germany, Greece, Ireland, Italy, Netherlands, United Kingdom and CERN . We have had six meetings so far-
20 Oct. 83 Strategies; Zander (D) 14/15 Nov. 83 Network-Layer; Bartlett (U.K.) 19/20 Dec. 83 Transport-Layer; Lenzini (I) 12/13 Jan. 84 Terminal-Access; Beyschlag (CERN) 23/24 Jan. 84 LAN-Operations; Jacobsen (DK) 13/14 Feb. 84 Session-Layer; Michiels (NL) 27/28 Feb. 84 Terminal-Access; Beyschlag (CERN)
The general idea was that the European harmonization activities should be directed to specify the choices to be made within existing international standards, to give the motivation for these choices and finally to give implementation guidelines. The following areas were covered in the EHA-
We intended to deal with the message handling system X400 and finally the file transfer FTAM when available as ISO standards. The results of this work, done by so many colleagues, were published in a set of papers known as the 'Common use Of ISO Standards' or COS. Four papers were produced. This is just to remind you of what we did in the past and it is my belief that the work influenced the actions of the Commission in the framework of the ESPRIT program in the Information Technology and Telecommunication Task Force in Brussels. You know that at that time CEN/CENELEC was appointed to deal with European standards and I believe the EHA was important to them.
We have to look and see what our role is in the environment we live in and what is our goal beyond the immediate ones already outlined. Currently we are dealing with networks of the classic packet switched type and the grand father of these is ARPA. What we are doing now in Europe is to install these networks and to try to achieve a convergence and harmonization of the different protocols in order to, hopefully, eliminate gateways. The developments are very different in our countries so I think we have try to do our best to come to a convergence, let us say, in the next 5 to 8 years. Before I go into more detail of what we should do let me tell you a little bit of the future, namely of the next decade when networking will be much more efficient and much more challenging than to day. This leads us into the broad band communications we will build and use in Europe.
In the Future we can expect to see very fast metropolitan area networks with connections into local and wide area networks. In Berlin we are considering such a network based on a mono modal fibre ring as the metropolitan area network. On this will be a variety of connections including 48K links to DATEX-P going up to fast couplings between super computers of 50M. The Berlin backbone is aimed at working at 250M. Other services of interest are connections to ISDN at 144K and high quality TV at between 34 and 144M. The network should connect to the German Bigfon telecommunications experiment. Currently we have plans for Bercom which connects various institutions in Berlin with PTT lines at 140M. These networks provide an overlay network to the local area networks. So I given you a view of the next decade with various types of high speed interconnected networks.
In Europe, some time ago, I made a proposal in Brussels to connect communications centers of excellence in Europe which would probably use satellites. This would give high speed connection to a wider geographical area. I mentioned this just to give you a view of how the future looks and how large a challenge it is to meet these needs and how harmonization is part of this work.
I have just spent a sabbatical half year considering these problems which is the reason why I have had time to think of what the aim of this workshop should be and what the outcome should be.
The ESPRIT program is very successfully operated by the ITTTF and has led to an intensive cooperation between European companies in the field of information and communications. The start of the RACE program will, hopefully, also be successful in the field of telecommunications. Certainly, CEN/CENELEC has been appointed to generate harmonization in launching and fostering European standards to be fed into the international standardization process and CEPT is providing guidelines and recommendations for the carrier functions of European PTTs. So far a high industrial and organizational potential has been activated and structured for large scale co operations in Europe. Finally, within this movement, an adequate participation of the academic community is missing. The dipole of industry and CEN/CENELEC should and must be widened to a triangle including the European scientific and academic potential. The cost 11 ter frame should be used. The most effective way to include the academic community in this strategic concept is to base it on national programs of OSI networking in different member states together with the Information Exchange System of the big 12 information technology companies. The key words are- interconnectability and interchangability. The EARN network demonstrates the usefulness of having a Europe or world wide communication network and the foundation of an EARN association last spring is a logical consequence of the remarkable EARN acceptance by users.
Future research and development in networking particularly, on high level protocols and those of broad band systems, need the participation of academic research and development work for non conformist and sophisticated users. For all these reasons it is absolutely necessary to bring together the national network program providers and their users into a foundation that I call JEANUP which is Joint European Association of Network Users and Providers which will fill the present gap and will provide a solid foundation for non profit making collaboration. [The association has now been provisionally called RARE - Re'seaux Academiques et de Recherche Europe'ens. It will be referred to as RARE in this text]. RARE could become a strong stimulus and motivate for a widened European collaboration if it is done in the right way. Some aims and goals are-
The triangle of cooperation is CEN/CENELEC, RARE and industry with the Commission, COST 11 and ITTTF in the center.
This week we must talk about these things. The core of these discussions between the national network programs is harmonization of networks. This must be fed back to the standardization bodies which is evident and vital. Of course, we provide the networking not for ourselves but for the users and that is the reason why users applications should play an important role to feed in new ideas for sophisticated services into the national networking programs. In addition research and development is needed soon into the way very fast networks of the future can be utilized, in particular the expected broad band communications networks.
Finally, in my view, any efforts which override national barriers which is international, like harmonized European national network programs, will have some political aspects. So in order to bring the ideas which will be born to fruition we need a critical mass of power formed from the large user groups like ECFA, the European Science Foundation, fusion research and climate research.
I hope that all of you will realize the tasks which are presented to you to be solved by you in a harmonized way. Let me say, and it is my strong belief, that if any group in this European community is providing real European cooperation then the national network providers are one of these most important groups. If you, colleagues and friends, will be able to install working networking based on these ideas to support the scientific European cooperation in the academic world or the industrial world then you build the bridges between the countries of our beloved Europe.
All workshop countries apart from Turkey and Yugoslavia have public X25 networks. In the case of Portugal this only allows calls out of the country. In several countries the services may only cover restricted areas of the country.
In most countries there is some form of academic network services. These are based on a wide variety of technologies and a wide variation as to the proportion of the community served. Most countries have based existing services on X25 with a variety of high level protocols. Most of these networks use the triple X protocols for terminal traffic. A few have private X25 networks but all, except the UK ,intended to use the public services in the future. The UK private network does have gateways to the public network. Leaving aside EARN, many countries have small pockets of proprietary networks, in particular DECNET. This is particularly strong in High Energy Physics and in Italy.
Most countries have or are putting together national academic networking plans. Without exception these are based on the use of ISO protocols. Germany has a well advanced plan followed by the UK who has the problem of migrating an existing X25 network to using ISO protocols.
It has been difficult to get figures on usage, charging, availability and what percentage of the community could use the networks. It is clear that one of the principle uses of the networks is for mail and it is fortunate that this is an area of ISO standardization which is well advanced.
Besides the public X25 networks there are a number of other international networks. For example the banks and airlines have extensive ones and in the research field there are the Euronet and EARN networks.
The principle private physical network of interest to academics is EARN. This was set up in the Autumn of 1983 and is financially supported by IBM generosity for 4 years. It uses the IBM RSCS protocols. There are about 100 nodes in Europe and 500 in its sister network BITNET in the USA. Several national networks, in particular SUNET, JANET and DFN intend to provide gateways to EARN. Although it only provides bulk services it is highly attractive as it is cheap and provides services to the USA.
NORDUNET is about to start and is a network connecting the Nordic countries. It will be based on the existing telecommunications services and national networks. It will require the harmonization of standards within the region.
DIANE used to use EURONET but now is almost entirely on the public networks. It provides information services from 50 hosts which contain in all 600 to 700 data bases. There are a number of international 'logical' networks which are groups of people who attempt to communicate over any networks they can. HEPNET was set up in 1981 by the High Energy Physics community and has been most successful in providing services with only modest developments. One of their principle activities has been the provision of gateways between various network technologies. These have usually been located at CERN. EUNET is somewhere between a physical and logical network. It is a network for mail between UNIX systems and tends to use dial up facilities. EUROMATH, like HEPNET, is a logical network, in this case, for passing information between mathematicians which has just started.
ESPRIT promotes five aims. Advances in micro electronics, advances in information processing, software technology, office systems and computer integrated manufacturing. To help achieve this it requires a communications infrastructure called IES (Information Exchange System). IES has been heavily involved with standards. It is trying to provide mail, conferencing, text preparation, information retrieval, integrated software development and graphics for the benefit of the ESPRIT participants. ESPRIT is a 5 year program but may well be extended. The first call for proposals was in 1984 and a second call was in 1985.
The Rose project, under IES, is aimed at the provision of OSI protocols mainly on UNIX. The CARLOS project is aimed at the production of equipment to allow anything to be connected to anything. The THORN project will provide name server and directory facilities.
IES hopes to use existing network infrastructure and not to build a separate network.
RACE (Research in Advanced Communications Environment) is aimed at providing a 'broad band' communications environment by 1995. The first part of the project is to develop the ISO model for broad band. The second part is to do the research and development required to develop the equipment needed. The topics for research are, high speed integrated circuits, high complexity integrated circuits, integrated opto electronics, broad band switches, passive optics, high bit rate components, long distance links, dedicated communications software and flat panel displays. The project will be subject to dynamic reassessment. It will be 50% CEC funded and cost 43M ECUs.
All PTTs were expected to migrate to X25 1984 but the time scale depended on the various plans of the PTTs, the plans of the manufacturers and also on the finance available. So far it is not clear what the inter working problems between X25 1980 and 1984 are.
Work was going on through CEPT to harmonize the actions of the PTTs. Hopefully the options implemented would be common to all European PTTs. Inter working problems were also being studied.
Three administrations were expected to have X25 1984 by 1987 and the rest would follow at various dates.
In some cases features mandatory in X25 1984 were available already such as fast select. Other features such as 'transit delay select' and 'throughput class selection' needed some research to find out what they meant and how they should be provided. Some administrations will not provide some features but all will provide enough to support the ISO network service.
The use of the X29 parameters had caused a lot of confusion. It had been recognized that successful services needed some recommendations on the use of the parameters and several groups had worked on the subject. The UK had produced the Green Book recommendations which unfortunately used features only available on PSS. ECFA had done a similar study which stayed within the standards. This work had eventually led to the harmonization COS document. Green Book also defined a means for operating triple X over non X25 networks and this work had been extended with a draft proposal for putting triple X over ISO session layer. Meanwhile CCITT 1984 had revised the standard and further complicated the protocol. In CCITT 1977 there were 12 X29 parameters, in 1980 18 parameters and in 1984 22 parameters.
The ECFA group had found that a large number of triple X implementations were non standard or substandard in that they used parameters in illegal ways or failed to provide some mandatory parameters at all. The way the PADs and hosts negotiated the parameter settings often left a lot to be desired. The provision of national parameters was another area of difficulty.
The ECFA and COS work takes the pragmatic view that one has to recognize that many implementations are poor and so devise recommendations to make interconnection as good as possible. Three modes of operation are recommended. The 'basic mode' deals with equipment that only provides the mandatory facilities or often less than that. 'Advanced mode' assumes that all the optional facilities are also available. The third mode is similar to the advanced mode and assumes that the 1984 facilities are available. It was recommended that only equipment capable of advanced mode should be purchased. Virtual Terminal Protocol (VPT) is not complete and an initial version is only going to be produced by leaving out many features where there have been difficulties. 6 classes have been defined- basic, forms, graphics, teletex, video and mixed. The basic class should be available soon as a draft standard.
There are still many issues to be resolved such as, 3270, other full screen services, performance of VTP, gateways between X29 and VTP, loosely coupled applications such as PCs and the future of quasi interactive protocols such as KERMIT.
A large number of file transfer protocols have been devised. In fact most networks have defined one. However ISO have almost finished the definition of FTAM which hopefully will be the last for some time.
There are, in fact 3 services required. The transfer of a single file. The transfer, examination and change of a file or file access. The creation and deletion of files or management. In addition services are not always between file store but may, for example, be to a job spool.
Most of the existing file transfer protocols only provided a subset of functions and often over a small range of lower level protocols. FTAM attempted to address all areas and, moreover use the lower level ISO protocols. There is some embarrassment that X400 uses different subsets of the session and presentation protocols. There are going to be problems in migrating from current protocols to the ISO one.
GIFT is one way of migrating. This project is aimed at providing a gateway between various file transfer systems. It currently or will soon deal with CERNET FTP, DECNET and Blue Book protocols. The project is realized on a VAX. It is hoped to incorporate UNINET at a later date. GIFT provides 'on the fly' file gatewaying. This method was selected due to the difficulty of knowing whether a transfer had been successful if staging were used.
It should be possible to integrate FTAM into the product and so produce a valuable migration aid.
There are currently at least six mail systems in Europe- COM, UUCP,ARPA, CSNET and EARN. It would clearly be desirable to have one system and to base it on X400. Unfortunately few X400 systems exist and any service needs to be harmonized and uniform.
X400 contains two parts. The User Agent (UA) which prepares messages and accepts them and the Message Transfer Agent (MTA) which ensures that messages are sent to and received from remote MTAs. There is considerable flexibility as to where and how the UAs and MTAs are located. It is possible to provide gateways between X400 and TELETEX and some administrations will do this. It is also possible to have gateways to TELEX.
A major problem is how the naming and addressing should be dealt with. X400 has naming domains and these have to be set up with some care for an organization such as the European academic community.
So far their are no commercial X400 implementations although a large number of manufacturers have developments under way. There is a version of X400 called EAN which has been produced at the University of British Columbia and it is on trial use at several sites in Europe.
In one year it should be possible to have an X400 node in each country. It would be necessary to have a directory service in each country. To do this some form of European cooperation would be needed to ensure that the system would provide services across Europe.
The rest of the world is principally what is happening in the USA where there are some 130 networks. Mail is one of the main networking activities and many gateways exist between the networks to provide services across networks.
USENET is dedicated to mail and has 2000 nodes worldwide. It uses the public telephone network and is run on a cooperative basis by the UNIX community. CSNET and MAILNET are also dominated by mail and have some hundreds of subscribers.
In most cases gateways work but they often provide poor failure notification facilities and directory services are also poor. In fact no gateways support any nameserver functions yet. There is little name registration to help users. In fact it is rare to find a good gateway as they are usually grafted on to machines providing other services.
Most of the networks do not allow transit traffic which is often needed because of the ad hoc topology of the networks.
An attempt is being made to set up 'Science NET' which is aims at connecting federally funded resources, in particular super computers.
Commercial networks are developing fast and are likely to provide X400 services soon. Some of these have international services and others are forming partnerships with other suppliers.
ISO in addendum to 7498, ECMA in 'OSI Directory Access Service and Protocol' and CCITT in study group VII have recognized the problems of naming addressing and directory services.
Since directories will have to be distributed the access to directories will have to be via networks and thus require a suitable protocol. The user will first access a local 'Directory Service Agent' (DSA) using a 'Directory Access Protocol'. If the DSA cannot satisfy the request then it will enquire of other DSAs using a 'Directory Server Protocol'. Work is going on to standardize the information which appears in directories. Several types of service are required such as Information services, white page services and yellow page services. Updating services are also needed.
To make such a scheme work there will have to be 'naming authorities' to ensure that names are unique and to provide conventions for naming and to manage it. To achieve this there will have to be a hierarchy of naming domains.
There are still problems to be solved with authentication and consistency. The Esprit Thorn project, and some projects within DFN are attempting to progress the provision of directory services.
The performance and failure reporting mechanisms in both private and public networks are poor. The PTTs complain that they suffer from lack of information from their equipment on national networks and the situation is worse with international links. In fact 5% to 7% of calls fail and this figure is rising. It appears that there are only primitive means for testing links and no dynamic reallocation on failure. The situation is made worse by the error messages on various networks being inconsistent.
It appears that there is little active monitoring of the public networks ether by the PTTs or customers. There is clearly a need to improve this situation and the academic community would do well to closely monitor the performance of the public networks and to have very close liaison with the PTTs.
The tariffs of all European PTTs are in the Eurodata Foundation Yearbook which, although expensive, is a mine of information. In most cases it is the charge per kilosegment which dominates. The charges across Europe vary considerable for the various services in each country. However there was no 'overall winner' as a country that had a cheap volume tariff might well have a high duration charge.
There are many questions that the community could ask the PTTs. Is the international surcharge reasonable? Why are there no discounts for night and week end use? Should there be bulk tariffs? Should the academic community have a special discount perhaps on account of possible collaboration with the PTTs?
An option that may be available is to base a European academic network on leased lines which should control costs and hopefully give a cheaper service. However this would require manpower, would incur large expenditure now and again, there would be legal problems and it may be difficult for small countries to participate. One could also ask why it had not been found necessary to have an academic telephone network?
The means of charging varied from institute to institute. If costs were small then charging could be part of general computing but it is possible for a user, possibly by accident, to incur high charges.
The principle problem seemed to be how to persuade the PTTs to charge a tariff that academics could afford.
ISO protocols start as working papers. They then become 'Draft Proposals' (DP). This is followed by 'Draft International Standard' (DIS) and they finally becomes an 'International Standard' (IS). Standards are usually quite stable by the time they reach DIS stage.
The current dates for ISO protocols are:-
The CEC became interested in standards in 1977. In 1979 the 'multi annual program' funded COST and CREST to further standards and undertake research and development for public purposes. In 1984 there was a Senior Officials Group on IT standards. In 1985 CEN/CENELEC became interested in the use and harmonization of standards.
In 1983/4 there had been an initiative to harmonize standards for academic use which had resulted in the COS documents. This work was taken up by SPAG and produced the GUS documents (Guide to the Use of Standards).
In 1984/5 there was CEN/CENELEC - CEPT harmonization.
CEN/CENELEC mechanism is that the CEC mandates a piece of work. After the technical work the document goes for public comment and then for ratification. It then becomes a Euro norm.
There are 5 priority areas. OSI level 1-3 and triple X. OSI level 4-7. Document transfer and messaging. Formal descriptive techniques. Programming languages, GKS and COBOL etc.
In 1985 there was a draft commitment directive on IT standardization following a directive on procurement. The work follows the following pattern. Determine the basis for standardization. Add precision. Develop new standard. Determine application of standard. Demonstrate developing norms. Promote the use of the IT standards. Management committee.
In fact 90% of the standards activities is in networking.
Currently JANET operates on the Coloured Book protocols. It is intended to migrate to ISO on an application by application basis rather than on a layer by layer basis. This will give fewer transitions. Various relays will be required to communicate between machines running old and new protocols.
In migrating to ISO network layer the network address will have to be rebuilt and it is intended to use the ISO DCC scheme and JANET is applying to the 'competent body' for numbers.
It is intended to apply pressure to manufacturers to provide all the network products rather than producing then in the community.
Currently all machines for general use in the community must provide Coloured Book protocols. This procurement policy has paid dividends in promoting JANET. Soon the procurement rules will allow suppliers to provide ISO products instead of Coloured Book ones and at an even later date ISO products will be mandatory.
The complete transition of JANET will take between 10 and 15 years.
There are now 70 machines in DFN which is based on the public X25 network. The services provided are triple X, file transfer and job transfer. The machines are VAX VMS, Siemens/BS2000, PDP11/RSX, ND100 and Siemens/BS3000.
DFN intends to use EAN. The migration to FTAM would take place in 1987 and to JTP in 1988. It is hopped to use manufacturers products wherever possible. One of the problems will be guaranteeing inter working. Another problem is the provision of stable user interfaces.
The European Academic Research Network (EARN) was set up in 1983/4. It uses the IBM RSCS protocols and operated over leased lines. The protocols limit it to file transfer and mail activities. There are about 100 machines connected and a further 500 machines on its sister network BITNET in the USE. The two networks are connected via two leased lines.
EARN is set up as an association set up under French law. It is run by a Board Of Directors, one from each country, who meet a few times each year. The international lines are paid for by IBM for 4 years but the network is controlled by the Board of Directors and IBM only contributes finance and advice.
Since EARN indulges in third party switching it requires licenses or permission from the PTTs to operate. In fact CEPT has taken a special interest in the network and has recommended that EARN must migrate to use ISO protocols and the public networks within some reasonable time scale. CEPT also recommended that the PTTs impose a volume tariff on traffic. In fact the tariffs imposed vary considerably with many countries imposing a zero volume charge and merely the leased line charge to the other extreme of imposing a tariff which is close as possible to the IPSS charges together with a fixed charge which approaches the leased line charge.
EARN has a technical group which meets from time to time to direct the technical development of the network. A small subgroup is studying how the network should migrate to ISO protocols. To this end they will soon be having discussions with IBM technical experts to see how best the migration should take place. There turns out to be several options from academic developments from Darmstadt and Salford University to ones based on IBM developments.
In Italy there is a policy of putting in gateways between the various networks. These are- X25, DECNET and EARN.
It is intended to put in two leased backbone connections. First- Rome, Pisa, Madrid and Florence. Second- Rome, Florence and CERN. The public networks would only be used for low volume traffic.
ISO standards would be used as they became available with performances comparable to the currently used ones.
48K lines are expected in a year with 128K via satellites available later.
NORDUNET starts this year and is a transition program based on the national networks. It is expected to take 4 years to produce a coordinated infrastructure providing a reliable service. There will be a harmonization of standard strategies and new value added services for end users. it has funding of 10M SEK.
The networks involved are:-
COST 11 ter follows COST 11 bis which followed COST 11. Ter like bis lacks the focus of a support network. Derek Barber was Director of COST 11 and Thomas Kalin director of COST 11 bis. The project leader for COST 11 ter has yet to be announced, although a recommendation has been made to the Commission by the Consultation Committee. The committee is responsible for managing COST 11. There is also a higher committee responsible for the 'Multi Annual Program' from which the budget is taken. The ter budget is small at 2M ECU plus small contribution from non Community countries for research and development. 1.4M ECU is allocated by COST 11 and .4M ECU is so far unallocated. Several of the projects are carried over from bis.
The countries involved are:- Belgium, Denmark, Spain, Finland, France, Switzerland, Italy, Great Britain, Ireland, Norway, Sweden, Germany and Yugoslavia.
The projects are:-
In my view, COST needs strengthening. ESPRIT, RACE and ALVEY are all exploitation and industry orientated and they are using the results of earlier basic research which has sometimes come from COST. This research activity is at a low level and the pendulum must swing back to provide more basic research oriented programmes, otherwise we shall have nothing left to exploit in industry.
There are three topics considered, the user needs, the advances in technology to meet the needs and the PTTs' plans. The users are the research community and those who wish to communicate with them over long distances. The ideas for future communications come from various conferences held recently including the ICCC 'Links for the Future' in 1984 and the COST 11 seminar in Rome.
The current user facilities are all limited by speed and better links will allow the following facilities to be offered- full screen graphics, large file transfers, fast fax, mixed text and graphics, slow and normal scan TV, audio, conferencing, broadcast, mobile and temporary working.
Current facilities, such as COM, are a good step forward as are X400. It is already cheaper to send an A4 page via X25 than by post but for 10 pages it is not true. On broadcast facilities the PTTs offer no useful services. Megabit per second working is attractive and ought to be important for everything as long as the tariffs are reasonable and the regulations for use reasonable. The universality of services is important for our community.
Technical improvements will offer 50K to 2M bps in next 10 years which will satisfy a large fraction of the academic users. However one must remember that a single LandSat picture contains 1 Gbit of information which needs a large bandwidth. There are 4 media, copper, radio, satellite and fibre optic, which will all still be in use in the year 2000. Currently the use of technology is 70% analog, 10% digital coaxial, 10% digital radio and 10% fibre optic. By 1988 this will have changed to 50% fibre optic, 30% coaxial, and 20% digital radio. Most data currently goes over analog links which often go over 64K PCM links multiplexed over 2, 8.4, 34, 139, and soon 565 Mbps channels. A new CCITT standards allows voice circuits over 32K channels. The PTTs see most of their revenue coming from voice in the future.
Although fibre optics are taking over many applications others such as TV will remain analog since a digital television picture needs 80Mhz whilst an analog one only takes 6Mhz.
Several new possibilities will become available with satellites. It will be possible to transmit from the satellite a narrow beam 100 miles wide which may allow some forms of switching. Another interesting possibility is to launch light aircraft which fly at high altitude and form a radio platform. These will act almost like a satellite but without the delay.
Fibre optics can now stretch for 30-40 Km without repeaters at 140Mbps. The next step should be 85 Km between repeaters at speeds of 565Mbps. In the laboratories distances of 200 Km at 0.5Gbps are possible. An interesting technique is to put several lasers on a single fibre thus giving a broad band capability. The PTTs are moving towards digital services and ISDN. BT expect to be fully digital by 1990 but subscriber links will not be digital until 2000 but they will be available for those who need them. It is difficult to see what the impact of ISDN will be and in fact the PTTs themselves do not know.
A basic problem is that the PTTs primary role is to provide speech services and digital ones form a small part of their business and it is grafted onto the speech network. Thus the measures of acceptability as regards error rates are not meaningful for digital applications
The PTTs still think in national terms and it should be a task of this workshop to persuade them otherwise.
The users want to do their normal work without spending any significant amount of time learning how to use networks. Since they are not and do not want to be network experts they are unclear how the networks will contribute to their work in the next 2, 5, or 10 years from now. It is therefore difficult to get an estimate of the users medium term needs. They always want money for their research needs and are thus reluctant to spend much on networking and thus need cheap services.
As well as new services the users need better versions of existing services.
The interactive services should provide more performance, in particular good full screen services together with editors are needed. There could be advantage in providing 'standard' methods of use on machines. There needs to be a solution to the problem of where to edit file- are they edited on the remote machine or file transferred to a local one first? Interactive graphics are needed which require a good bandwidth.
Scientists would now like to control their experiments remotely to avoid wasting time and money in travel.
Electronic mail must be easier to use and generally available. The users should not need to know about the details of the networks but only the name of the recipient and possibly the site he is at. Mail should be able to deal with formatted text, graphics and other information so that it can be used as a simple form of file transfer which does not require a knowledge of the remote file store. Although EAN is a step in the right direction, implementation of X400 on a wide range of machines as commercial products are needed. The need for mail across Europe is urgent.
There seems little need for digitized voice yet.
Computer conferencing is widely used and popular in some communities. It does seem that it suffers from the problem that users have to access a remote machine to use it and they would rather conference entries could come to their home machine. These systems also suffer from not providing a wide character set. Tele or video conferencing is not used much and there seems little need for it. Its disadvantage is that the participants all have to be present at the same time and for a limited period. This technology does not seem to encourage 'coffee' style discussions which are important. The problem may be in the limitations of the equipment rather than the basic facility.
File transfer falls into two groups. First there are small file such as programs, data samples and graphics of less than 5M bytes. Second there are the large files such as experimental data and digitized pictures of 20 to 200 M bytes. In both cases the users want reliable, cheap and responsive services although large transfers may well be less urgent and can be transferred at off peak times when tariffs are low. As soon as it is possible to transfer large files the service will be heavily used as long as tariffs are reasonable.
Services to transfer text which is capable of re editing and re formatting are needed as a lot of traffic will be directed towards the production of documents. As yet there are no standards apart from the de facto IBM Script one.
Users seem happy with the current remote batch processing facilities based on the 'Red Book' or IBM protocols. Theses services are heavily used.
There are problems with the concept of the scientific workstation. Currently these are small machines with alphanumeric or graphics screens and slow links into remote machines. In the future the computing load needs to be shared between the work station and the central machine so that the tasks are done on the machines best suited to the jobs. A medium speed reliable network is needed for this service but it is unclear what it should be. Can the digital telephone services be used for example? It may be appropriate to try and distribute most of the computing load to machines located in various places on a site.
Scientists often like to work out of hours or from home. The principle problem is to have fast links to the central site and the current 300 or 1200 BPS services are too slow and unreliable.
The 1200 to 64000 bps traffic can be expected to increase in particular on international links within Europe. Higher speeds of 512 to 2000 K bps are needed now and equipment to deal with these speeds is also needed. To deal with such speeds high speed digital terrestrial links, satellite services and X21 connections are needed. The PTT services should be cheaper, less rigid on the services provided.
The applications provided should be to ISO standards and these standards are needed quickly. The interpretation of the standards should be done carefully to ensure inter working across Europe is possible. To achieve this effective European coordination is needed. Pressure is needed on manufacturers to provide these products. In one or two years time products should not be bought from suppliers who do not provide suitable communications products. Gateways are needed to communicate with networks which use less standard products.
The main restriction on traffic is the high tariffs and this conflicts with the need to put more money into the science itself. A special tariff for academic traffic would be very welcome. In particular the removal of the premium on international European traffic would be helpful.
Users needs are growing and the principle requirements are for mail, file transfer facilities on high bandwidth, reliable networks which follow standards. The networks must be coordinated across Europe. The tariffs of the PTTs are currently too high and need to be reduced if the user needs are to be met.
Present:- A Patel Ireland M Walsh Ireland M Wells UK F Ros Spain K Neggers Netherlands B Mahon CEC J Harms Suisse A Kundig Suisse M Paul Austria M Hine CERN B Carpenter CERN F Fluckiger CERN E Valente Italy J Prevost France U Beyschlag CERN P Villemoes Denmark B Carlson Sweden P Kirstein ARPA liaison B Butscher Germany J Hutton UK P Van Binst Belgium K Ullmann Germany N Newman CEC J Renuart CEC K Zander Germany D Barber COST 11 P Linington Chairman P Bryant Secretary
The purpose of the European networkshop was to investigate the possibilities for providing good international networking services to the academic and research communities in Europe. The purpose of this meeting was to develop a proposal for continuing support of this aim to be put before the full meeting of the workshop.
There was unanimous support for the view that common and good networking services across Europe depended on the development and implementation of the ISO protocols. It was recognized that the successful implementation of these protocols depended on the harmonization of the options in the protocols selected as well as harmonization of items where decisions were needed such as naming and addressing.
It was proposed that any early activity should be directed to exploiting low speed technology up to 64K bps. Interest will be likely to shift to higher speed technologies as these develop and as lower speeds become well established and require less attention.
It was proposed that the responsibility was to provide networking services although some development would be involved where the goal was immediately related to this.
It was proposed that the networking services should if possible be common to the academic community and to the industrial research community.
It was proposed to form an association. It was proposed that membership of the association should be on a 'country' basis rather than being an association of 'networks' or of 'interest groups'. It was proposed that membership, at least initially, should be confined to Europe.
The association required to be connected with some existing organization in order to support. The CEC was considered inappropriate as it would have excluded, for example, the Nordic countries and this was unacceptable to the meeting. It was proposed that COST would be a suitable grouping of countries.
It was proposed that CERN should be a member in the light of its special international character, its strong requirements for networking and its networking expertise.
The COST grouping excluded Israel. It was proposed to create an observer status for countries which would allow then to participate but possible not to receive funding or possible having to pay a subscription.
It was suggested that the association may need a small permanent secretariat to undertake organizational activities and secretarial services.
Professor Zander proposed the aims of the association to be:-
M Hine proposed three work items:-
It was considered important that services should be provided as soon as possible to create momentum and show results. The Migration to ISO protocols, which is also vital, will take longer to accomplish.
The proposed aims and work items were agreed.
It was unclear where finance would come from and how much was required. It was certain that early support would be needed for international working groups. It is to be hoped that the use of electronic mail and conferencing systems could reduce costs.
There was a suggestion that only modest finance would be required for development work as several national programs are already very active and the harmonization of the work was important rather than initiating new activities.
It was agreed to produce a financial plan in the near future and to seek suitable sources of funding. The CEC asked that some outline plan be developed as soon as possible for presentation to an ESPRIT meeting in mid June and that a detailed plan should follow.
It was agreed that the association should be directed by a group with academic representatives form each country. Prior to the establishment of the association it was suggested that the workshop organizing committee should draw up a proposal for immediate work items. It was unclear how industry, PTTs, CEC or other groups should relate to the association.
It was agreed important that the object of the association was to foster networking and not the provision of a network. The association did not want to be seen in any way as a competitor to the PTTs. It was also vital to maintain good relations with CEPT as well as to foster good relations between national networking groups and their PTTs.
It was clear that the CEC wished to see industry taking advantage of the networking provided but it was unclear how the relationship between the two communities should start and develop. It was agreed that such relationships were important and to be welcomed. It was reported that after a few initial difficulties industrials and academics had worked well together in ESPRIT.
It was hoped that the association would be a powerful influence on manufacturers to provide the network components that were required. This must be achieved by close consultation and possibly by preferentially purchasing from companies providing products to the harmonized standards.
It was agreed that work should be started immediately and not wait for the creating of the association which could take a long time.
Eight possible priority areas were identified:-
It was unclear what methods should be used in each area and it was agreed to ask the organizing committee to discuss this the next day.
It was resolved to put the above plan to the full meeting and seek their comments.
Further to discussions between CERN and Rutherford we would like to request an X25 'Coloured Book' connection into the CERN VM/CMS service for the benefit of the UK HEP community.
The service will provide:-
Benefits expected:-
Hardware requirement:-
The software to be mounted will be provided by Rutherford and is:-
No modification are required to CP or any other software. However generation options are required to allow any number of wrong passwords to be quoted by the NIFTP machine.
The modems/line drivers are for further discussion.
The CERN bisync port will be maintained by CERN.
Installation and maintenance of the VMNCP and NIFTP software is adequately documented, and has been installed by seven other VM sites with little or no reference back to RAL. CERN will install the software and integrate the service machines into the normal operation of the VM service. In the event of problems, CERN operators and systems programmers will provide first level support. Unresolved problems will be referred to RAL Service Line or RAL Software Support, as appropriate. The software has been operational at Rutherford under VM/CMS for three years and earlier versions for eight years. The software is also mounted on HEP machines at Glasgow, Imperial College, Birmingham and Liverpool. It is also mounted at a few other sites. It has been found that the product is reliable and requires little maintenance. In general any problems can usually be resolved from Rutherford.
With the introduction of the 4705 and COMPRO software at Rutherford it is expected that the connection between the Rutherford IBM and X25 equipment will migrate to use HDLC. After a suitable period of confidence running at Rutherford it may well be advantageous to convert the CERN connection to conform. This should improve the bandwidth and move further towards ISO standards. It will also allow the link to be moved to CERN supported X25 switches should this be thought desirable.
Your agreement to proceed with this project as soon as possible is requested.
Dear
I have produced the attached proposal for the EARN Board of Directors meeting on the 24 June. I am afraid that it has been done in some haste and I expect it is littered with mistakes and misunderstandings concerning the facts and also what we agreed. I would be most grateful if you would take a look at it and let me know if there are any changes you would like me to make. Unfortunately time is short if I am to meet deadlines for producing an agreed document by the meeting and so I must ask you to act quickly. I will also attempt to send this to you by EARN.
May I take this opportunity of thanking you for attending the recent meeting which I think was a grate success.
Yours sincerely
Paul Bryant.
Dear John,
I attach a document entitled 'The status of JNT-recommended protocol implementations'. This shows the current status of all the coloured book products.
The UK has an X25 private academic network called JANET. Each university has (or should have) a local area network connecting to it. Most of these are also X25 but these are a small number of Cambridge Ring and Ether net ones. We have a Joint Network Team located here at Rutherford which ensures that only coloured book products are used. We do not allow the purchase of university computers unless they provide these facilities. This has proved a successful policy- you will see from the sheet that most machines now have a more or less full set of protocols.
We recognize that the coloured book protocols are only standard in our community. It is also evident that there is a need for open networking round the European academics and also to other continents. The only solution we can see is to use a non proprietary set of protocols and the only set that any one can agree on is the ISO ones. It is to be regretted that these have taken so long to be defined. Currently only FTAM is needed to complete a working set which could replace our coloured book. I ignore the Red Book which is job transfer as this is not yet well established.
Thus we are committed to migrate to ISO in a controlled and well ordered way to prevent upsetting the users. In fact most European countries are also committed to migrating to ISO and there is now a strong move to ensure that the protocol options we select are harmonized across Europe.
IBM is financing the international lines for EARN. In the UK we are having a gateway between EARN and JANET at Rutherford. Other countries have allowed to EARN exist and provide a partial national network and Germany with 60 connections is an example. Of course, EARN only provides file transfer and mail. In addition we have had to get permission from the PTTs to run it and they are insisting that it migrates to use ISO protocols and public networks in the next 4 years. So it will have a short life and possibly a gay one.
The coloured book protocols are mainly confined to the UK. Ireland has also adopted them and will migrate with us. This is very convenient as there are very close ties between the two countries which foster the idea of close network cooperation. Several high energy physics sites are also using then in particular on VAX computers. This is because the VAX product is good and also these sites want to network over the public networks and this is the only way as DECNET has a number of draw backs as you probably know. I would estimate about 20 European sites have it. In the rest of the world there are isolated sites which again are mainly HEP. I guess about 4 in America and then there is Japan. There is one Australian site running the DEC10 product. I think they are on 5053003410 or 505273720000.
I guess that if you were to adopt coloured books then you would have to tie yourselves closely to the UK and migrate to ISO as we do.
The chap who came to work with us for 6 months was:- Frank March, DSIR National Computer Network, Computer Research Section, Physics & Engineering Laboratory, Department of Scientific & Industrial Research, Private Bag, Lower Hutt, New Zealand. Tel 666 919 WN.
He knows all about Coloured Books and I would suggest you contact him.
I enclose a document I wrote a long time ago which gives some of the history of JANET. It was called SERC net and ICF network from time to time. In fact now there are approaching 1000 machines on it including those on the local area networks attached.
I also enclose the summary of the recent European Networkshop which gives you a flavor of what is happening. I have included the address list of participants which you might find useful. As you will see the Association we hope to form will be strictly European but I am hopeful that we can include other countries as observers which means that you get no voting rights and no financial assistance. I am sure that any country which shares the European desire to follow ISO will be welcome even though observer status was really invented to keep Israel happy.
Our gateway to the public packet switched network is 23422351919169. This will get you a message asking for your user name and password which you ignore. In fact you have to type the name of the machine you want to access. An interesting name is .STATUS which allows you to see what the network is up to and to find out what other names there are. Most names are 4 characters with the first two being the site the third the machine type and the fourth and ambiguity remover. We have stopped having guest accounts because of the amount of misuse but you can try .RLVC with user name PEB and password PAUL. It is one of my many accounts but one I rarely use. You can send me mail with:-
POST/TO=BRYANT/AT=RLGB
This drives Grey Book mail.
I hope this is useful to you.
Best wishes
Paul Bryant.
The 'Committed Strategic Plan' (CSP) is better described as a discussion document for further stages of the ELAN project. It is comments on a number of ideas as to how ELAN could develop. It does not define the objectives of the work nor is it specific about the details of the products to be used or the services to be provided at any point.
I would have like to have seen the document stating:-
To further clarify my feeling I produce below a skeleton document which is illustrative rather than correct.
Introduction.
The first and second phases of ELAN has resulted in the installation of three CSMA/CD networks. One of the networks is for development and the other two provide a service and are interconnected via a leased telephone circuit. There is a gateway to the public switched X25 data network which provides interactive facilities. The hardware is to ISO standards. The services provided use a number of software products some of which do not follow ISO standard.
Requirements.
ELAN is provided to :-
The priorities of the Task Force for services are:-
High priority requirements should be met as soon as possible and may involve temporary equipment and methods pending development of products following ISO standards. Medium priority requirements should be met with hardware following ISO standards but the software used may not meet standards but must be made to do so in due course. Low priority requirements should only be met with equipment meeting ISO standards.
Strategy.
The strategy is to select and install products which conform to ISO protocol standards and provide services required by the Task Force. The network is expected to be completely ISO within two years.
Tactics.
Few products are currently available which provide ISO protocols and a service which is of use to the Task Force. There are a number a products which conform in part to the standards and which will conform in the future.
In order to provide any demonstration or early service, partially conforming products must be used.
Where the user demand is strong non standard products will be provided but only if the products can migrate to full ISO standards or will be replaced by such products as they become available.
As products following standards become available the reluctance to permit products which do not conform to continue to be connected or introduced will increase.
Remember that the above is only illustrative!
At the conclusion of phase I the situation is that local area networks conforming to the ISO CSMA/CD standard has been installed and commission. There is a minor problem in that some 'taps' do not conform to the standard but these will be or may have already been replaced. This work looks to be very satisfactory.
Phase II provided a service over the LAN installed in phase I.
The basic services are provided through a 'NIM'. This is a Z80 microcomputer which multiplexes four V24 interfaces onto the LAN and thence allows connections to other NIMs which each have four more interfaces. Some interfaces are connected to terminal equipment and other to ports into computers. Thus the user's can first communicate with the NIM and establish a path to a selected service. This type of equipment is very popular and available from a range of manufacturers keen to provide a 'turn key' installation. The protocols used by each manufacturer are all in general different. It provides a good 'excuse' to install a LAN without the protocol and interworking worries but with limited services. All it really provides is for each terminal to get at each and every computer.
A second type of NIM, which consists of the same hardware but different code, provides a bridge between two CSMA/CD LANs over a synchronous connection.
A third type of NIM provides an X25 PAD function and this gives access to the public switched data network.
Apart from the CSMA/CD hardware and X25 functions the NIM does not conform to ISO protocols.
All other equipment currently connects to the LAN via the NIMs. Thus the only standard here is V24 which is not very interesting.
In comment- the route chosen is an excellent way of getting visible progress quickly and does not prejudice the future too much. As a long term solution it is not satisfactory as it does not allow easy and convenient computer to computer services required by mail, file transfer and like services. In addition it requires a lot of asynchronous channels and PADs with their channel limitations and 'birds nests' of wires.
Phase II (according to the 'Road Map') installed a set of services conforming in part or whole to standards. The standards quoted are V24, X3, X28, ISO DP 8802/2, ISO DP 8802/3, ISO 646, SIC-SI. These are to provide mail/document transfer facilities and access to external services. The omission of X29 is presumably a drafting error as this is the fundamental protocol for interactive services.
In fact, the use of triple X is confined to the X25 gateway where it is essential if any service over the public data network is to be provided. The rest of the interactive services use non standard protocols.
The mail/document service is a PC to PC or word processor to word processor function based on a simple protocol over asynchronous connections via the LAN. This is a well tried and proven scheme which although a manual activity is very effective. As a long term solution it is not good as it is a labour intensive and a more automatic system is to be preferred which can be provided using ISO protocols.
Phase II of ELAN seems to have met its target although the methods used are a little pedestrian for a set of companies at the forefront of network development. They are very similar to facilities put together very effectively in many places using pragmatic methods with the hope that it is possible to migrate to standards in time. In fact many sites have based their initial developments on the use of a 'standard' low level and boxes which provides asynchronous connections through the network. These same sites, at least in the UK, are working towards replacing such temporary methods. For example, at SERCs Daresbury (P Kummer) laboratory they installed an Ungerman Bass ethernet, which is similar to the system with the TASK Force, and this has now got a lot of other equipment connected which is moving towards the use of better protocols. Bath University (J Thomas) has tried to avoid the PAD solution and has some interesting projects with ICL and GEC to provide kit to ISO standards.
The principle work item is to upgrade the network to use the ISO transport service according to the appropriate SPAG profiles.
It is unclear how this is to be achieved. It clearly requires a new set of products which may be new hardware and software and/or software upgrades to the current systems. Since this phase is scheduled to start now and last for a year the products needed must be in a fairly advanced state. The ELAN team had clearly not examined this in detail although it is clear that manufacturers must have products which are almost available now if the time scale is to be met. In addition it seems likely that the NIM could be upgraded to conform to the relevant profile and this seems a strong probability. It seems curious that the software for the NIM, which is important for the project, has been contracted out to a small UK company. At this early stage one would have thought that the companies would have liked to do the work themselves for the experience they would have gained.
Where there should have been some information is on the products already installed such as the NIM. Other questions are- will the UNIX systems be supplied with direct connections to the LAN rather than by PAD connections? Similarly will the PCs and word processors also have direct connections. Clearly there is a very wide range of options ranging from merely upgrading the NIMs and probably providing no further functionality, to providing transport service on many types of equipment and possibly new equipment. If transport service is to be provided then it is interesting to ask what protocols will be provided above transport.
My guess is that the intention is to upgrade the NIMs fairly rapidly and to possibly take advantage of UNIX or PC developments which could mature towards the end of the phase.
Phase III intends to provide TELEX and TELETEX gateways. Again there is no indication of how this should be achieved. It would be relatively easy to connect TELEX or TELETEX 'boxes' via a few asynchronous interfaces to NIMs so that the user merely accesses them as he would a remote host. It would be relatively difficult to move TELEX and TELETEX data to and from users mail boxes, say on a PC or UNIX system, automatically. The former route would be merely to provide a service to the Task Force and have little bearing on standards whist the second would have everything to do with standards. My suspicious nature guesses that they mean to take the easy option. However, let us not be too critical as such a simple service would be highly useful and would give, let us say, 80% satisfaction over the 10% satisfaction with manual TELEX services.
The intention to provide a 'standard mail delivery service', which seems to be X400, is very exciting and the most important aspect of phase III. The implications are large and if they achieve the aim then they would indeed be praised. The statements imply that some or all systems will have a direct connection to the LAN offering transport service. It also implies that these systems will have X400 mounted over then together with user interfaces to the mail boxes. Since they expect to interchange mail with external services then that implies that there will be a transport gateway to the public X25 network.
Currently there is, I believe only one or two pilot X400 implementations and these are designed mainly for use over X400 and not LANs. Moreover these products are not from the consortium. Now I know that X400 is high on their list of priorities and thus I have confidence that the products will appear but I would not be surprised if they were a little late.
A problem that may well be to provide a gateway between transport class 0, X25, X400 and transport class 4, X400 on the LAN. I know that products are being developed but I have some doubts as to how satisfactory they will be. I worry over the addressing problems and the miss match of the facilities between the transport classes. This is a comment on standards and not ELAN.
The conclusion I reach is that some of the objectives, such as a TELEX server, appear easy and are done to meet the task forces requirements and give confidence that improvements are taking place- no bad thing. Other objectives are exciting and difficult such as X400. A cursory reading of this section gives no indication of which of the objectives are in line with the pursuit of standards and which are expedients for quick services.
The study areas are good but again the objectives of following standards and providing services is unclear. The 'common directory' is claimed to be a 'standards following activity' and should no doubt be tied in with the THORN project. The printer, file, archiving and scanning servers to the best of my knowledge have no standards activities attached to them and will no doubt rely on some underlying file transfer mechanism such as FTAM. These are clearly aimed at providing services to the Task Force.
Since this phase does not begin until end 1986 it is perhaps unkind to be too critical. There are, however, a few puzzling questions which have a bearing on phase III.
The claim is that a session service (ISO session service?) will be made available to support applications profiles, in particular mail. This is puzzling since X400 had already been provided and so must have a session service to allow it to talk to remote sites. In the next section the mail service is expected to be upgraded to conform to SPAG profiles. Again if it was not already conforming how was it able to communicate? I make the guess that X400 will turn out to be a 'mess' with different people using different flavours of the standard. Thus, I interpret the section to mean that the team will strive to make the products talk to as many people as possible and to align with any harmonization going on. Again- a laudable objective. I worry that the text seems to suggest that the project intends to migrate to ISO by 'layer replacement' rather than 'application replacement'. The UK academics rejected layer replacement as being a very difficult option.
Hopefully the provision of FTAM in the time scale will be possible. By the time we get to this stage one would hope that most if not all equipment would be directly connected to the LAN and that the NIMs would have been phased out or re-deployed as gateways or something.
This is so far away that it can only be taken as a statement of intent to want to continue the project with whatever comes along- and I am pleased to see it. The protocols needed to achieve some of the objectives are only in their initial stages of definition. It is also fairly clear that the equipment needed will be considerably more powerful than that now employed. For example the digitized hand written document is likely to take at least an order of magnitude more storage than one in ASCII. If you are moving data around in this form' is a 10Mbps LAN fast enough? None the less, I am pleased that they are considering these things.
This section is contentious and is influenced by the activities of the UK academics. If their views are correct then you should be worried.
ECMA, in their wisdom, have decided that transport class 4 should be used over a LAN. This view is reflected in the relevant GUS profiles.
The alternative to class 4 is to run X25 with class 0 or perhaps 2.
ECMA have elected to use class 4 as it avoids the complexity of X25. X25 also imposes heavier overheads as it introduces an added level of acknowledgment. In fact, X25 is really designed for unreliable networks and thus has appropriate error recovery to achieve reliability.
The UK academic community have decided to use X25 over LANs. The reason for this is that it simplifies gateways and also allows the same X25 code to be used on the WAN and LAN. The academics also feel that with a 10Mbps LAN the overheads are unimportant.
My personal view takes the argument a step further. There is a class of protocols, such as Newcastle Connection for UNIX and DR Net for the IBM PC, which are so called 'light weight protocols' based on connectionless principles. It turns out that these protocols are very easy to implement over a LAN and provide services which are ideal for distributed systems. Sets of PCs or UNIX systems can make very effective distributed systems where users can easily access disks on other machines or make use of servers as if they were part of his machine. Regretfully, there are as yet no ISO protocols on this area but I believe that these will be developed. Thus, I see light weight connectionless protocols being used very effectively on a LAN in an organization such as the Task Force. Such a LAN would have one or two machines which would act as X25 servers for the other machines and would thus want to pass X25 data across the LAN to a gateway to the WAN. The adoption of X25 on the LAN is a first step towards such a world and it is a happy accident that as yet we do not really need the LAN bandwidth. By the time that we do I would hope that the light weight protocols were well established and would obviate the need for extensive X25 traffic on the LAN. In fact, one is putting together two separate logical networks on the same physical network- connectionless and X25- both to ISO standards. In fact, a single physical interface to a machine can run both 'stacks' of protocols at the same time. This presents no logical problems.
ECMA, in deciding on class 4, are attempting to make LAN traffic cheaper on the grounds, I suppose, that there will be a lot of connection style traffic. This, I believe, is a wrong assumption.
Thus I believe that the adoption of class 4 transport on a LAN is an error. However I could well be wrong.
I am told that at the last ISO LAN meeting the use of X25 over a LAN was recommended rather than class 4.
This section goes through the document making comments in order.
Page 5. The document claims the 'Further guidance has been sought for the commission [on user requirements]'. I am surprised that the team did not have discussions with the Task Force prior to producing the paper or that they were not already well aware of them. After all they are in the same building and have been working together for some time.
Page 6. The first paragraph rightly draws attention to the conflict between the user needs and the desire to respect standards. The ELAN project recognizes this and this has led to the phased approach. I do not see how the phased approach follows but on the other hand I do not have the original proposal.
There is a claim that system software used is 'product line'. It is not clear to me that this has been true in the first two stages as the NIM software seems to be specifically for ELAN and the document transfer is an 'off the shelf' back to back transfer mechanism not part of the suppliers product line. I may be wrong. However, I think the requirement or desire to use products which any customer can buy is important or else the benefit of ELAN as a 'shop window' will be lost.
The concept of 'islands' of equipment each running incompatible protocols over the same LAN describes very well what is happening. I have difficulty seeing how these islands can be progressively interconnected step by step. I see a much more likely progression being the elimination of an island and its replacement by 'mainland' ISO as a sudden occurrence. For example, it is difficult to see how the current back to back connection of word processors for document transfer being gently step by step replaced by something like FTAM or ODA. There seems no graceful migration path. The number of islands must be kept as small as possible for as each new island is put in it will present yet another migration problem.
The 'illustration' of the step by step migration in terms of X400 and ODA is interesting. Clearly X400 and ODA work will come to fruition and be used in ELAN. But- how do you organize a step by step migration to it as I strongly suspect that X400 and ODA will arrive from the manufacturer as a 'take it or leave it' package with no pre releases of partially X400 or ODA products. In this paragraph there is a suggestion that ODA is in some way related to X400. This is not my understanding. In fact the text is ambiguous. On the other hand I do not see X400 as a vehicle for the transfer of formattable text in general.
I like the idea that the products to be displayed at SICOB and HANOVER being put on ELAN as a continuing demonstration in a more realistic environment. It will be very interesting to see what is demonstrated as this will be a good guide as to the equipment the four will propose for ELAN. Since the SICOB is in 1986 the equipment to be displayed must be in a late stage of development.
The achievements to date are nicely defined in the second paragraph but hide the real nature of the technology which I described earlier. Incidentally I never did find out what EPPA and EMA stood for.
The third paragraph is important and implies the use of X400 on various equipments. I strongly suspect that this must be interpreted as X400 on the UNIX machines. I would be very surprised but also delighted to see it on the PCs and word processors which the passage implies. This work, if restricted to UNIX systems, will only be interesting if there are multiple UNIX systems or a gateway allowing X400 traffic to other sites is provided. A small point- I do not see why 'multi column' documents cannot be dealt with. It smells as if they have some specific product in mind- what is it?
On page 8 I applaud the enhanced document handling but I find the idea of 'handwriting' in a processable form to be a long way off. Further off than mixed text and graphics in fact. A comment I marvel at is 'The relative ease with which [in ODA] the graphics facilities can be added to the text.. which has not generally been found in proprietary systems'. One wonders how many non proprietary systems will have this feature implemented in the foreseeable future.
Paragraph 3 page 8 is important in that it defines an actual product to be produced and used in ELAN. It would be interesting to see a specification of the product which could answer a lot of my questions.
Page 8 section 3.3 is all very correct. However much of the work of the 'wheel of achievement' is now well understood in that many standards are well defined and the SPAG profiles are known. Once the wheel has 'turned' then I see future developments being of a different character. I do not see difficult migrations but rather improved products with more reliability and more functionality but working within the original standards. I am not at all sure that future standards activity will be of the same character as the current ones. I see new protocols being in the applications areas and attempting not to change the underlying protocols. An exception to this may be the development of protocols to deal with new technology such as satellite links, say.
Page 10 discusses SPAG. As was mentioned by Prof Zander at the Luxembourg networkshop - the current work is based on a cooperation between industry and standards bodies. This has to be extended to include the academics. This may well have an effect on the SPAG profiles. I worry that the academics have not been involved in the SPAG work and in some cases have been working on alternatives in ignorance of SPAG. Some further harmonization may be needed.
The direction SPAG has taken of defining profiles may encourage the development of non reusable levels. Thus a presentation layer developed for X400 maybe unusable with FTAM. Also the manufacturer may not provide interfaces except to the application and this would prevent the customer developing applications products. It can already be seen that IBM ISO products appear to adopt this approach and has already prevented the use of their X25 product supporting coloured book products.
A further worry is that there are several companies which are producing ISO products who are unaware of SPAG and are working to other principles. An important job will be to make SPAG as well known as ISO or there will be a danger of the development of non interworking islands of products which none the conform to SPAG.
In 3.6 I applaud the statement that eventually any products to be put on ELAN will have gone through a validation process. Since it seems that these tools will be available in march 1986 it implies that only full ISO products will be attached after that date. I hope this is true.
Section 4.1 gives a rather poor indication of the Task Forces requirements and is merely a list of random items. The diagram shows every conceivable interconnection between entities and as such is not very helpful is indicating which paths will be implemented in ELAN. One can be sure that not all of them will be there and moreover several of them are of no interest. I would have thought that the aim was to minimize the variety and number of physical interconnection whilst ensuring that all logical connections are possible. To draw an analogy- there is little point in having twenty different taps in the bath room if the only object is to get water.
Section 4.2 which attempts to refine the requirements is also curious in that several of the topics they mention which effect design are already cast in concrete- for example, the topology of the building. So whist the text is all very correct it would have been nice to see some conclusions drawn from the actual ELAN case.
Section 5 defines in detail the SPAG profiles to be used and this is a mechanical extraction of the profiles from the SPAG recommendations knowing the ELAN topology.
Paragraph two states that the SPAG profiles will be used by SPAG members. I hope that they will be used a lot wider than that or the exercise will have been a failure. Certainly I, as a customer, want to buy my equipment from a wider manufacturer base than SPAG members.
In clarification it seems that there is really no distinction between the WAN and PLAN. They share a common address space and therefore there is no gateway between them.
In profile 1 there are two possibilities shown. It should be noted that the first option suffers from not having a non null internet layer which effectively prevents calls using this scheme passing through gateways. I do not understand the comment that 'where [internet functionality] is needed it can be supplied by the gateway'. The only case where this is possible is on an interactive call when the user can stop at the gateway and provide the additional addressing information needed. I am glad to see the second profile will be increasingly used. I am very surprised to see the first scheme being defined as a SPAG profile at all.
In class 3 we see the interesting problem of how you relay between transport class 0/2 and 4. A small point is that no mention is made of the problems of using X25 1984 with its address facility.
Class 7 illustrates the problem of X29 over a LAN. Regretfully X29 uses the 'Q' bit for indicating command messages and this lives inside the X25 protocol. Thus if X29 is used over a non X25 network there is a problem of what to do with the Q bit. That is what PAD adaption is all about. Several candidates have been proposed for PAD adaption which is why it appears with no ISO number. A rather heavyweight idea from the UK Department of Industry is to place X29 over ISO session layer. This has been rejected by the UK academics in that there is difficulty in providing the functionality needed- in particular how you deal with a 'break' and with 'expedited' is very messy. The basic problem is that X29 breaks all the ISO layer structure but is none the less a fact of life which has to be lived with. The easiest solution is to run X25 over the LAN which solves the problem at a stroke but SPAG do not like that idea.
After very careful study the document does make a lot of sense and the ideas are all technically and practically sound with a few exceptions. What is annoying is that it is so difficult to extract the information into an understandable form. I believe the document could be cut down to half the size with no loss of information. Also a rearrangement of the information into a more logical order could be very advantageous.
As I said at the beginning- the document needs a clear statement of the requirements , strategy and tactics. The tactics need to be the bulk of the document and also the part of the document which will change with time as products and standards mature.
The reference sites which you may like to approach for an opinion of the GEC 6300 series are:-
Rutherford Lab. Bob Day Cambridge University Graham Titmus Edinburgh University Dave Mercer