Paul Bryant's Networking Correspondence
Dear Dr. Kington,
The Davos Institute proved to be a valuable and enjoyable experience. I came away having made many new friends and made many valuable contacts which are sure to prove valuable in the future. The talks and discussions gave a rare opportunity to delve deeply into the two topics which are of interest to me just now- networks and personal computers. I am most grateful to have had the opportunity to participate in such a fruitful and well organized event.
Having enjoyed it so much I feel rather guilty in introducing the sordid matter of money. However I understand that I am entitled to a small honorarium for giving a presentation and Lassi Hyvarinen informs me that I should be receiving this from you.
Yours sincerely
Paul Bryant.
EarnTech/1/84 List of names, addresses and telephone
numbers of the EARN Board of Directors
and IBM advisors. P Bryant
EarnTech/2/84 Batch Simple Mail Transfer Protocol E A Crosswell
EarnTech/3/84 BITSERV Anon
EarnTech/4/84 Letter on naming M Hebgen
EarnTech/5/84 EARN naming and gateways P Bryant
EarnTech/6/84 A simple guide to the JNT name
registration scheme. P Bryant
EarnTech/7/84 Letter to M Hebgen- names and addresses
on EARN P Bryant
EarnTech/8/84 Naming convention for node names on
EARN Germany
EarnTech/9/84 IBM (CMS) as a FTP 'P' end P Girard
EarnTech/10/84*EARN Technical Documentation Index P Bryant
EarnTech/11/84 NETSERV B Pasch
EarnTech/12/84 Mailing Facilities at Tel-Aviv
University NOS
EarnTech/13/84 A CYBER-IBM connection O Comay
EarnTech/14/84 RSCSNET FILTER B Pasch
EarnTech/15/84 Lettre d'Information EARN-FR J L Delhaye
EarnTech/16/84 Results of survey P Bryant
EarnTech/17/84 EARN Technical Documentation Index P Bryant
Documents marked * are obsolete.
This was probably THE crucial meeting. Most of the meeting was taken up with PTT and CEPT problems. It now appears that EARN has reached a good understanding with CEPT and there is a telex defining this understanding. Some PTTs have how provided circuits and the rest are expected to follow shortly. BT, or rather DTI, has not yet agreed. The EARN organization has now taken some more definite shape with various responsibilities being taken by members. IBM is taking no executive role in EARN but is none the less taking an active interest in encouraging development.
The BOD is now most ably chaired by Dennis Jennings from University College Dublin. The IBM site hosting the meeting provides a secretary. There are a number of working groups whose chairmen together with Jennings provide a 'cabinet' which attempts to clarify topics and present recommendations and options to the BOD. This 'cabinet' consists of:-
On this occasion the 'cabinet' had a pre meeting the previous day which significantly reduced the discussion at the main meeting.
Hargen Hultzsch had written to Thomas Hubner of CEPT with a statement of his understanding of the terms under which CEPT would not oppose EARN. Hubner had responded with a telex which restated the terms in rather more flexible and advantageous terms (for EARN). The statements made are:-
The third item does not imply that the volume charge need be non zero. The statement is one of principle and has been added since other organizations with leased circuits have caused CEPT some concern.
In the light of this telex it is now believed that the only impediments are the national PTTs.
Relations with PTTs are confused. For example, the French regulatory authority have agreed to a connection but the installation side will not put it in. Hopefully such problems are just bureaucratic delays. In general most PTTs have agreed to install. BT has not yet agreed. The BT delay is due to the privatization of BT which has meant that the Regulatory Affairs department has moved from BT to DTI and the process of obtaining a license has had to recommence with a set of new faces. However it seems likely that a license will be forthcoming in the near future.
Most PTTs have agreed to levy a zero volume tariff and it is hoped that BT will follow suit.
Ireland. No connections. Plan to connect 20 machines on 7 sites.
France. Connections to 4 sites. Plan to connect 20/25 machines on 4 sites.
CERN. Connections to 2 machines.
Switzerland. Plan to connect to 10 machines.
UK. Plan to connect to 1 site and 1 machine.
Belgium. Plan to connect to 16 machines.
Denmark. Early planning stage.
Germany. Connections to 54 machines. Plan to connect to 100.
Italy. connections to 12 machines on 3 sites.
Spain. Connections to 3 machines on 3 sites. Plan to connect 7 machines on 7 sites.
Israel. Connections to 3 machines.
Sweden. Plan to connect from 3 to 5 machines on 3 sites.
Finland. Early planning stage.
NOTE - THE ABOVE FIGURES ARE INACCURATE.
There were considerable difficulties with this for the UK. The problems were:-
It was decided to call the charter a 'document of understanding' to prevent any legal problems.
This was a demo of a network display centre on a 3279 model 3B. It was rather similar to the JANET network display. It showed a map of Europe and the lines between sites. These lines turned different colours depending on whether the line was up, down or carrying traffic. It allowed a certain amount of traffic monitoring.
There are now 74 sites and 220 machines connected. There are 5 gateways to other networks. A network information centre has been set up. It is expected that many of the facilities of BITNET will be imported to EARN.
After a short discussion on the move to international standards it was agreed to produce a survey paper on the options which would be drafted by David Lord and Paul Bryant. It was felt that a distinction should be drawn between the use of standards and the use of public networks. In addition a distinction should be drawn between the use of X25 and the higher level protocols.
IBM was asked for a statement on its support of ISO but this was declined on the grounds that the public statements already made should suffice.
It was suggested that publicity should concentrate on contacting the end users with announcements in various learned (and not so learned) journals. Regretfully one country had produced an EARN logo using the IBM logo type face which was not considered a wise move. A number of draft logos were on display. Perhaps there will be EARN T shirts soon!
Birgitta Carlson gave a presentation. It was agreed that EARN needed conferencing for two purposes. Firstly for the BOD themselves and secondly for the users of EARN. It was agreed to use QZ COM for the BOD and to commission a study for a general facility. Ira Fouchs and Birgitta will do this.
The COMPRO software makes the 3705 run like a drain. Peter- please note.
Heidleburg are attempting IBM to IBM PROFS connections over X25. Keith- please note.
Heidleburg are attempting RSCS over X25 but this product is unlikely to see the light of day.
Germany has at least 12 full time people working on EARN with a lot of IBM support.
Thank you for replying to my request for information. I am circulating the useful documents I have received and I am reproducing other information in this document.
End user documents - Montpellier. User documents - Madrid. EARN user guide - Germany. EARN pocket reference summary - Germany. EARN facilities guide - Germany. A description of the Israeli section of EARN - Israel.
Israel are checking a CDC product named NJE as a replacement to TIELINE, since TIELINE does not support interactive messages.
Dear Mike,
Further to our telephone call I have collected a few documents that may be of interest. Unfortunately, since the network is still in its formative stage there is no good description of its state at any time.
The documents I enclose are:-
Minutes of last but one Board of Directors meeting. Please treat as semi confidential as I would not like them to get to the press. It will give you some idea of what we are talking about.
Letter from Hargen Hultzsch to CEPT giving his understanding of the CEPT position.
Telex from Huebner giving, what the BOD considers, to be the official CEPT position. If indeed that is their position then the BOD is happy to accept it.
The draft EARN charter which gives the restrictions that the users will operate under.
The last document is a letter from our minister, Kenneth Baker, in response to one by David Hartley asking him to expedite matters. In fact I think the situation has changed in that other PTTs are now happy to allow EARN to exist.
The current situation is that international connection exist between Rome- New York, Rome-Madrid and Rome- Israel. Switzerland- Rome and Switzerland- France lines should be available soon and only technical and bureaucratic matters are in the way. Germany now has an operational network within Germany of 55 machines and expects this to rise to 100 machines. Sweden, Denmark and Norway are still in the early planning stage. The Ireland- UK line is waiting UK approval as is the Switzerland- UK link. I am none too sure of the above facts and there are gaps in my knowledge so do not take them as gospel.
It appears that most PTTs are happy to impose a zero volume tariff which maintains the CEPT principle but allows the academics to have a limit on expenditure for the time being. Germany is intending to impose a volume tariff but only for calls emanating from dial up connections- a curious condition.
Finally, may I ask you to see if you can speed up the granting of a license or at least allow us to proceed on the understanding that a license will be forthcoming.
I will be most happy to give you any further information you require or to visit you to discuss the matter further.
Best wishes,
Yours sincerely
Paul Bryant.
Access to the computers at the Rutherford Appleton Laboratory is mainly via a variety of complex communications networks. The computers themselves are used by a few thousand scientists and administrators who demand a wide variety of services. As the networks have grown their development and maintenance have become very important and so we are looking for further staff in this important area. The computers range from large IBM and Amdahl machines through VAX and GEC mini computers to IBM PCs. Normally we try to use international communications standards or those defined for use within the academic community. These standards include X25, Ethernet and Cambridge Ring hardware. The software follows the 'Coloured Book' standards but ISO standards are being actively developed. We are involved with both the academic network JANET as well as the proposed European Academic Research Network EARN and so have close contacts with other UK and European groups. We are looking for graduates for permanent or fixed term appointments. If you join us you will probably be able to work in any of the network areas. Although communications experience is desirable a strong interest in the subject and a willingness to lean are more important. These appointments offer a unique opportunity to work in an exciting and fast developing area where you will have the satisfaction of providing facilities for demanding customers.
As you know, Informatics group is welcome to attend NDM meetings and I have issued NOTE 3 as an NDM paper. I intend discussing it at the next meeting. Without prejudice, several of the questions you ask can be answered if my understanding of the Ethernet is correct. The paragraph numbers follow those of NOTE 3.
May I now add a few extra comments:-
Experience with Cambridge Ring has taught me a lot. The prime lesson is that we should provide no service or even plan a service before we have fully investigated all aspects of the technology and we have a reasonable range of proven products on which to launch a service. We must ensure that any service fits in with the network strategy which is broadly to follow JNT approved standards. We must ensure that a service network connects to the major machines and that we can province service on demand to a reasonable proportion on the site who may require then.
These comments should not be taken to mean that bits of network cannot be installed for special purposes where appropriate. Thus if you want to have some Informatics Division private network round your UNIX machines based on Newcastle connection then that is fine but it must be recognized that such a service cannot be regarded as a laboratory standard service and requests to provide connections to equipment not in the Division will probably be refused. Indeed the proposed IBM - VAX connection for HEP is just such a special connection as was the proposed SNS - IBM Hyperchannel connection. I would certainly not demand that you should follow the protocols on ether of these networks or ask you to join them.
Gunadhi will be undertaking some work on Ethernet so I am hopeful that a good degree of cooperation will be possible.
I understand that JNT may well be supporting the X25 option over Ethernet. We must follow their deliberations with care. If they do take this path then I would expect them to pressure the other academic sites, including Edinburgh, to follow suit. It is my intention to follow JNT regardless of the decision they take. It is easy to say that since our initial project will not be in that area!
I trust these comments will aid our deliberations at NDM.
Digital communications between computers and terminals have been developing since the late 1960s. There have been a large variety of methods or protocols used and the complexity of these have increased as the demands of the applications used has risen. This has led to machines from different sources being unable to communicate with each other. At the same time the users have increasingly wanted to pass information between a variety of machines and to have the freedom to purchase equipment from a variety of sources. The larger manufacturers have developed sophisticated communications architectures to meet the needs of their users and to match the applications they provide. The smaller suppliers have attempted to capture a small corner of the market such as the interconnection a few work processors from a small range of manufacturers.
Clearly the diversity of products has prevented the easy interconnection of heterogeneous products. there is no technical reason why there should not be a single set of protocols to meet most requirements. There are many reasons why such a goal is difficult to achieve. A few reasons for the difficulty are- manufacturers have vested interests in 'locking' their customers into a particular set of standards- it is difficult to reach technical agreement on the specifications as various experts have conflicting ideas- it is difficult to persuade manufacturers to adopt these standards unless they can be pursued that it will increase their profits.
Now that a set of non proprietary standards are becoming available from ISO the question arises as to how they can be implemented and popularized so that it become inconceivable that a manufacturer should not supply then in just the same way that it is inconceivable that FORTRAN should not be supplied. It turns out that even when a standard has been defined further work is needed to define suitable interworking subsets to ensure heterogeneous products can be interconnected. It also turns out that manufacturers show some reluctance to adopt the protocols before a full set have been defined. The manufacturer is faced with the problem that missing protocols can prevent the use of these standards for a large subset of existing applications. For example, IBM 3270 are not catered for and the file access methods offered by DECNET are absent.
ESPRIT, and similar programs, can play a part in the early development of implementations of the ISO protocols. However, just like a manufacturer, ESPRIT has the problem of wanting to provide services at an early date and so faces the dilemma of non standard methods now or standard methods in the future.
The talk will describe the development of the ISO protocols and explore the problems of producing networks based on them.
Dear David,
I enclose a copy of the FDDI (Fibre Optic Token Ring 100M LAN) standard from ANSI. I do not think it is the latest but I think it is sufficiently stable for this document to be fairly accurate.
I also enclose the proposal which was submitted to DTI. Since this was contributed to by companies I would ask you to regard it as 'Commercial in Confidence'
The final document is some comments on the FDDI meeting which gives a flavor of what is going on.
Perhaps if an ESPRIT proposal is a possibility CERN may like to be involved. We would need to find a non UK company to collaborate.
I hope you find the documents interesting.
Yours sincerely
Paul Bryant.
Dear Nick,
I enclose a copy of the FDDI (Fibre Optic Token Ring 100M LAN) standard from ANSI. I do not think it is the latest but I think it is sufficiently stable for this document to be fairly accurate.
I also enclose the proposal which was submitted to DTI. Since this was contributed to by companies I would ask you to regard it as 'Commercial in Confidence'
The final document is some comments on the FDDI meeting which gives a flavor of what is going on.
Perhaps if an ESPRIT proposal is a possibility CERN may like to be involved. We would need to find a non UK company to collaborate.
I hope you find the documents interesting.
Yours sincerely
Paul Bryant.
Dear Derek,
I enclose a copy of the FDDI (Fibre Optic Token Ring 100M LAN) standard from ANSI. I do not think it is the latest but I think it is sufficiently stable for this document to be fairly accurate.
I also enclose the proposal which was submitted to DTI. Since this was contributed to by companies I would ask you to regard it as 'Commercial in Confidence'
The final document is some comments on the FDDI meeting which gives a flavor of what is going on.
As I mentioned in Brussels, I have half a mind to see if I can get some Esprit funding as it seems to be in line with some of the things they are up to.
For the record, I am intending to get the CSS code for an IBM PC from Prof. Schindler to see how it goes and it would be valuable to run tests between a couple of places in the UK as well as to Brussels. When I get some more definitive information I will come back and talk to you about funding.
I am also intending to look at the Louis Pouzin MSH code but I am not sure how to mount it. We do not have a C compiler on our GEC computers and might have to use some other machine.
Lastly I am hoping to try out the MSH code from British Columbia, again with something of a spirit of adventure. My principle problem will be to find staff to look at all these things.
I hope you find the documents interesting.
Yours sincerely
Paul Bryant.
Dear Francisco,
Thank you for your interest in the UK Academic Network - JANET.
I have put together a few documents which I hope you will find useful.
The first is a paper I gave at an IBM conference which is a good description of how the network developed.
The second document is a paper I wrote for Electronics and Power. It is called the ICF network but it is the same as SERCnet and JANET and it is interesting to compare the two papers to see how it has developed.
The next paper was written by Mike Wells and gives some advice to those wishing to connect.
AI/69 is a paper given at Networkshop and gives some interesting technical information.
Throughout the documents you will see references to the Coloured Book protocols. AB/AE/49 gives the current state of the protocols on a variety of machines. As you will see, most 'interesting' machines have implementations of the protocols. I have not included the protocol documents themselves as they are large and unless you are interested it would be a waste of postage. If you would like them I can send them to you.
I enclose two copies of Network News which will refer to further documents I can supply on demand.
Although the low level protocols (X25 and triple X) are international standards, the higher levels are not. We are committed to move to the ISO higher levels in the next five years and this should be very interesting but difficult.
Interestingly, Ireland are adopting the Coloured Books for their academic network as they provide an 'off the shelf' network scheme. They will therefore take part in the transition to ISO that we will be involved with. Although I would not wish to influence what happens in Spain it is a possible solution to providing a working network over X25.
I hope you find the documents useful and I will be happy to supply any further information.
I should add that each year we hold a conference called 'Networkshop' at which the various aspects of academic networking, as well as local area problems, are discussed. At the last conference we had several representatives from other countries and I am sure a representative from Spain would be most welcome- it is cheap as it is in a university.
Yours sincerely
Paul Bryant.
I went to the Technical Week as I have been involved with the ESPRIT Information Exchange Scheme (IES) and had been asked to give a talk on standards in relation to the various networks in Europe.
The first 3 days which I attended were a conference and the next 2 days were a workshop which I did not due to pressure of work. About 600 people attended. There was one standard which was observed universally- the use of English. Many well known faces were seen- Derek Barber, Brian Randall, Gordon Scarrot, plus many familiar faces from ICL- Plessey- GEC. From the continent many of my networking 'muckers' were on hand such as Louis Pouzin, Andre Dathene, Dennis Jennings, prof Zander and the good professor Helms. Notable by their absence was the IBM contingent.
The first morning was a waste of time as it consisted of a plenary session on rallying the troops by Carpentier, Roberts (GEC) and Syrbe. The afternoon livened up with Cadiou of the Commission giving us a survey of the strengths and weaknesses of Europe. EEC seems better in conventional areas such as optics, software, CAD, telecoms, applications, aerospace, cars (do my ears deceive me), banking and insurance. We are bad on robots and real high tech things. Thus Esprit is to put things right. He went on about how we must 'raise level of interface' (if only we knew what that was), must integrate, must get stuck into enabling technology and so on. You could hear Beethoven's ninth being hummed by the assembled multitudes. There were now 450 applications for the 50% funded projects in the next phase. Each project must have partners from at least two countries. An interesting area where few projects had been received was software technology and computer aided manufacture.
Brandin from Stanford talked about the glories of the USA and Japanese programs. Such terms as 'redefining the technology base' came easy to his lips as did the DARPA budget of $50M for 1984 rising to $780 in 1988. In fact the talk was very interesting as he contrasted the ways of doing things in the various countries and warned that you cannot transplant culture or rather the way various nations work best. The fact that Japanese workers stayed in the same place for life while Americans 'job hop' is not something one should try and change as both ways of working have their advantages and disadvantages. He pointed out that most Japanese research was well published although unfortunately only a very small proportion got translated. The Japanese have taken the trouble to learn English to have far better access to US research than we have to theirs.
Mackintosh of Mackintosh International talked about the European IT industry and the results of a survey. His first conclusion was that it was almost impossible to compare the countries as the statistics were ether unknown or presented in radically different ways. It was evident that France was spending more than any other nation with Germany and UK fighting it out for second place. We sunk to fourth place when it was expresses as a per capita expenditure. He recommended- better statistics, more coordination, more cooperation near the market place and he thought Esprit was a good step.
This broke into 6 parallel sessions. I gave the one on welding technology a 'miss' and went to the Information Exchange System. We first were treated to a presentation by Elie on the implementation of the OSI protocols in EIS. This is a vast project I am associated with. It starts, or started, with messing about with 'cu' and 'uucp' on UNIX in a rather poor way and has since spawned a very grandiose plan of evolution towards a full OSI system primarily on UNIX. I cannot say I like the way it is being done very much as they are tending to do it in a very piece meal way by elevating things like 'uucp' over X25 and other rather expensive short term expedients. Their work is currently at variance with the Harmonization ideas so something will have to be done. They are talking about using Class 4 transport over their Ethernets instead of X25 and they intend using class 2 and 3 over X25 instead of 0 and 2. However I guess it will get sorted out. One of my sorrows is that they are concentrating so much on UNIX rather than the protocols as only 21% of the machines are UNIX. There is almost no work on VAX/VMS or IBM which are common in the Community. A good point is that the code will be available to members of the ESPRIT club effectively free. Although the project has a lot I would disagree with it is a positive endorsement of ISO and should yield a lot of implementation experience and a lot of working products- or should I say interworking products. I almost felt the spirit of our early X25 work in SERC was abroad.
Bendayan gave a talk on UNIX United and in particular how they intend to run it over session layer and wide area networks. In the discussion he seemed to be a bit woolly on how he would know when to open and close a session link. He intends to multiplex all traffic between 2 machines over a single connection. It seems each application would have to in some way inform the session layer of its needs and the session layer would keep track of the requests and only close the session when all requirements had been met. This seems non trivial to me. There was a lot of discussion on the best place to put communications code on UNIX and most people seemed to be waiting for the future versions of UNIX which had suitable Kernel facilities for putting the session layer and below in the Kernel. The current front end processors did not seem to be all that popular on cost, inconvenience and 'they don't seem to work to well' grounds. However there did seem a lot of interest in multi processor UNIX systems with the other processors (on the same bus) being dedicated to some of the communications tasks but still running UNIX like code.
Schindler of Berlin University presented the CSS-conference support system. This is something I was trying to get Alvey interested in. Basically it proposes that the committee work of ESPRIT should be conducted using TELETEX terminals of various sorts. He has TELETEX code for UNIX and PC DOS. The UNIX code is 200 ECUs which are the units of EEC funding and worth 50p). The PC version is 1000 pounds and the added cost is due to a board being needed. The code is 50,000 lines of C. It looks to be quite an exciting project and the manuals for it are 6 in number and very detailed (technical not user sort). We would do well to follow this work. My worry is that TELETEX will be overtaken my MHS which was the subject of the next talk.
Ulf Beyschlag talked about MHS or X400 protocols which are the CCITT mail protocols. These are the ISO version of Grey Book Mail so are very important. He approached the job in a humorous way and gave a delightful talk. He asked the question as to whether MHS was the answer to a maidens prayer- the maiden being Miss Europe. He noted that all the mail systems left a lot to be desired in that one could not pass mail to other systems which had mail without a lot of difficulty. People were getting fed up with the complexity of mail gateways and all the horrors of the current systems. Naturally he claimed that MHS was the answer particularly as the PTTs were supporting it and public mail systems would be based on it. Unlike TELETEX, MHS says little about the user interface and the presentations aspects which TELETEX goes to town on. It does talk a lot about how mail is passed round networks. An interesting piece of work (not reported in the conference) has been done by IFIP under the guidance of Danthine of Liege to define many aspects of the user interface - in particular how people should be 'named and addressed'. I believe that MHS will be very important and we should take steps to become familiar with it.
On Tuesday afternoon the oxygen level in the crowded window-less, air conditioning-less room was getting low whereas the alcohol level was high which resulted in me missing some of the finer points of the presentations. I apologies to Egmond for not remembering a word about his 'choices for an LCS independent interface'. Rey spoke about local area integrated optical networks which while interesting was really surveying the possibilities. He was interested in the possibilities of some PABX facilities and other services all on cables using some broadband type technology. It was a bit futuristic. Danthine gave an exciting talk on 'Broad site wideband local networks' which was a misnamed talk. He talked about some very interesting results on LAN performance. (It is amazing how people manage to get their latest ideas into their talks regardless of the title). An exciting result is that the IBM token ring is not very good as the priority mechanism limits the performance to far below maximum and that the ring without it had a far better performance. As long as guaranteed performance was not demanded the priority-less token ring came out very well. He showed some very interesting graphs that I hope will be published. He mainly concentrated on packet sizes of 1000 and 5000 bits. The Cambridge ring showed up exceeding well. This was a point I raised in the discussion and found that the reason was that the mini packet size was 1000 or 5000 bits which is a little unrealistic. He is wiring the large Liege site with a backbone network using wide band technology.
It started with a plenary session and Hunke outlined what would happen in the parallel sessions.
I went to the ESPRIT IES session and had to listen to Bryant giving one of his usual high quality well rounded presentations. In this case he took standards as his topic- what had been happening and how this effected the academic and research networks. He make a few coarse comments on the IES project and was pleased to see it generally supporting standards. He was very worried about the EARN network which he understood would be using the IBM RSCS protocols and was scheduled to link 200 machines my Christmas. This would make it one of the largest academic networks in Europe. He welcomed its agreement with CEPT to move to ISO protocols.
Pouzin followed on with a discussion of his work on interconnecting mail systems with a system which translated all mail to a canonical form called X400 (surprise). His system currently gets connected in a variety of ways and polls for mail. It uses X25 as well as dial up. The work is supported by ESPRIT and we can get hold of it for roughly zero pounds. It is in C and runs on UNIX.
Newman went further into IES and how he saw it developing. He was mainly interested in how the users could use it and the benefits it would bring. He also covered the success of the COM system in Dublin which now has a few hundred members. Indeed, I am a member and I find a lot of my GEC and ICL friends there whereas the QZ COM tends to gather my academic friends. The world is indeed getting smaller.
Wednesday afternoon was a plenary session. Every parallel session had had a rapporteur and the plenary was taken up with reports. This was quite a surprising session as it gave a completely different view of ESPRIT where communications sank to a small corner of the scene. Robot welders and micro circuit technology surfaced.
The impression is that the pilot ESPRIT projects have been very successful and have produced good technical results. It has been found that the various European companies can work effectively together. In fact they welcome working together. I was left with the impression that ESPRIT was the type of think that the Community was all about rather than the more public butter mountains. ESPRIT has certainly taught me a lot about European industry and what the European academics are up to in my field.
In summary it turned out to be an excellent conference.
I brought back a few documents which may be borrowed-
Six volumes of CSS papers.
Call for proposals for IES
Electronic Office Systems- Standards. Schindler.
Implementation of OSI in IES- Many authors.
UNIX united aspects of IES. Bendayan.
Future Generation Computer Systems- Vol 1 No 1- North Holland (library should get it).
Many pamphlets on Eurocom, the ESPRIT UNIX mail system in Hurst, Diane, Pouzin's mail gateway and more.
Dear Mike,
Thanks for your note. Regretfully I do not see how the 'Applicable Branch System' license helps in this situation. However, let me first answer your questions.
I attach a diagram showing a simplified picture of our communications arrangements and where equipment is situated. The three large boxes represent what is in the building, what is on the site and what is in the UK. PADs are 'Packet Assembler Disassembler' which are really terminal concentrators. The IBM 3726 and the Amdahl 4705 are communications processors specific to the IBM and IBM 'look alike' machines. The double lines indicate a number of lines.
DESY and CERN are the High Energy Physics research laboratories in Germany and Switzerland and are linked to Rutherford to enable the UK scientists to use the computers at Rutherford from the remote sites. In addition it allows the UK scientists when at Rutherford to access their machines at DESY and CERN. There are substantial networks at both CERN and DESY but traffic to and from the UK is regulated to ensure that 'third party traffic' regulations are not breached.
The EARN part of the network, which is boxed in red and is not installed, connects to the IBM 3726. Traffic is possible between Ireland, Rutherford and CERN. Via CERN traffic can pass to other parts of EARN and to BITNET in the USA. Traffic can pass between JANET and EARN.
As you will see the network is quite complicated even in its simplified form but I think the diagram covers most possible types of connection.
All the traffic is data. The traffic across EARN consists of files and of electronic mail. JANET also carries interactive traffic.
There is access from the public telecommunications systems. There is dial up for terminals but such traffic cannot pass over EARN for technical reasons. There is access from and to PSS and again interactive access from PSS and thence over EARN is not possible whereas file and mail traffic is.
Ownership of all the equipment on the Rutherford site rests with SERC which is a branch of DES. Equipment on JANET is owned by a variety of organizations principally universities, research councils and some polytechnics. As far as I know there are no commercial organizations connected to Rutherford or JANET and no commercial traffic is allowed.
I have had a look at the document you sent me on 'Applicable Branch Systems'. This does not address the problem of third party switching which is the basis of the problem. As you will see, we indulge is considerable network activities which leave the site to both UK and European sites and also use a variety of telecommunications facilities. These are allowed as we take care to fall in line with BT technical regulations and DTI regulatory controls.
I regretfully believe that there is no option except the issue of a special license along the lines of the CEPT statement which I sent you. Certainly I explored various options with Chris Hobson and we found nothing. The only faint possibility was a 'Value Added' license but this does not reflect the spirit of the network.
I am wondering whether I should visit you and perhaps meet Alister Macdonald to expedite matters. I am sure you can understand that my European colleagues, particularly in Ireland, as well as the potential users are pressing me to provide a service. Perhaps you would 'phone me or I will 'phone you later this week to decide how we can proceed.
I am most grateful for the trouble you are taking to resolve this difficult problem.
Yours sincerely
Paul Bryant.
Circulate:- P F Linnington, J J Thresher, C J Pavelin for consideration
Circulate:- J Barlow, D Botterill, P Bryant, P Girard, N Gunadhi, J Thompson
The HEP Division requires a connection between their VAX/780/750 computers and the IBM main frame in Computing Division for the transfer of large amounts of data held in archival storage on the IBM. This requirement is related now to:-
In the longer term, higher speed connections to CERN/DESY are envisaged to which a LAN would be linked.
The 370/E emulators provide additional CPU power for routine processing of large amounts of data held on archival storage on the IBMs so that for good flexibility, it is essential to have high speed connections to the Computer Center where this storage will be held. In a present application, one 370/E processor can analyse a full 6250 bpi tape of data in about 5 hours (0.6 x IBM168 equivalent power). With a small cluster of such processors, one should aim for transfer rates of less than 30 minutes per tape corresponding to 500 to 1000 K bps. At present, the emulators are attached to the VAX by a 10 M bps Ethernet and a high speed network to the IBM should have at least the same order and preferably a higher capacity to allow for expansion.
Four technologies have been considered. There appears to be no other technology which is sufficiently well developed or freely available to expect that a service can be provided in the near future. The four technologies are X25, Hyperchannel, Cambridge Ring and Ethernet.
X25 technology is currently limited to about 50Kbits/sec. This data rate is considered too low. 2Mbits/sec interfaces are being developed but it seems unlikely that an interface of this speed will be available on the IBM (or VAX ?) computers in the near future. The speed of 2Mbits/sec is only just sufficient for the applications and when protocol overheads are taken into consideration the speed is likely to be insufficient.
Hyperchannel equipment gives a data rate of 50Mbit/sec. It had been intended to use this technology for SNS using the NETEX protocols but the monthly rental of the software of 700 pounds per month for the IBM and 600 pounds per month for the VAX was considered to be too high so the project has been shelved for the time being. This situation is likely to be reviewed. The software was given a trial for two months and was shown to work although the speed was disappointing. The use of the Netex protocols would have had the disadvantage of being out of line with the protocols being developed for other technologies in the academic community. This disadvantage would have shown itself in only allowing the network to be used for file transfers between the IBM and the VAX computers rather than allowing it to be used for general purpose traffic possibly to and from any computers in the community. It would be possible to develop and use accepted protocols over this network. Such a development would be about the same magnitude of work as similar protocols over other technologies and in fact much of such code could be common to several technologies. The development of Hyperchannel in this way has the disadvantage that the protocols would only be of interest to Rutherford or the small number of Hyperchannel sites. Hyperchannel equipment exists for the IBM and for VAX. The VAX equipment allows 4 computers to be attached. The hardware is very expensive at 30,000 pounds per box.
Cambridge Ring has been in use in the Laboratory for some time and therefore its characteristics are well known. The basic data rate is 10Mbits/sec but this is reduced to at most 1.3Mbits/sec due to low level protocol considerations inherent in the technology. The fairly crude nature of the IBM interface limits performance to 100K bps but this could be increased by the use of more sophisticated equipment. Use of CR82 protocols, which are the only standards available, will also depress performance considerably. Attempts are being made to redefine the Ring protocols to bring it more into line with ISO standards. This will provide alternative lighter weight protocols which may be more appropriate to the applications. It is uncertain whether this move will be successful and thus use of the technology is a risk. Cambridge Ring equipment with software is just becoming available for the VAX from Acorn and, it is understood, an alternative product will be available from Camtec. Hardware without software has been available from Logica for some time but is inherently more expensive to run as the other offerings incorporate processors on the Ring boards. It is known how to produce an IBM interface using the Auscom IBM interface and a Logica Ring Q-Bus interface. It had been intended to develop such an interface at Rutherford but the incentive disappeared with the very late delivery of other ring equipment and lack of staff. Thus the Ring has a number of uncertainties and a barely adequate performance which makes it unattractive.
Ethernet is now becoming an ISO standard as are the protocols to operate over it. Equipment is becoming available and its price is likely to drop. The network provides a data rate of 10Mbit/sec. However, the contention nature of the technology means that continuous rates cannot be maintained above 80% or severe degradation can occur. Equipment is already purchased for VAX computers in connection with the 370E and a HEP group has considerable experience with it. Spare VAX Ethernet equipment is available in Computing Division which can be used on a temporary basis. It is known how an IBM interface can be provided using the Auscom IBM interface with an Interlan Q-Bus ethernet board and this equipment exists. There are still a number a protocol problems and the experts within the academic community and DTI are actively attempting to solve these in the next few months. The technology has the advantage that the light-weight protocols needed by some applications can be run concurrently with the more traditional protocols which give interconnection with other networks and computers. Thus the technology does appear to have a long term potential and it will be popular. This means that protocol products are likely to be available from elsewhere for some applications.
There are two types of protocol available over networks these being connection and connection-less.
Connection protocols have been used over the X25 network exclusively. In using these protocols a data connection is established between the two applications which then exchange data. When they have finished the interaction the call is cleared. This is analogous to a telephone call. The protocols ensure that the connection is error free and this caused considerable overhead in mechanisms for detecting and correcting faults. These protocols are used where open systems interconnection is important.
In a connection-less network an application can send data to any other application. There is no guarantee that the data is received or that the sender is informed if data is lost. This means that protocol overheads are very low but the difficulty is that the application has to provide any error detection and correction needed. These types of protocols are of most use in distributed systems (or data collection) where the machines interconnected are 'closed' and the applications can deal with errors.
Both types of protocol are needed and it is advantageous if the same network can support both types of protocol concurrently. Such a local area network could then support the traffic envisaged by the HEP Division as well as being used for the type of network traffic supported by JANET.
The CERN 'Data-gram Definition Group' are considering the use of the connection-less ISO protocols (ref 1). Thus this proposal closely follows their work. This should have the effect of reducing costs as hopefully some of the CERN work should be available for this project. ESONI are also supporting the CERN work.
The most promising technology is currently Ethernet. It is a standard, it can support all types of protocol. It provides a sufficiently high data rate for current activities. It is popular and equipment exists.
A sufficiency of protocols are now defined to allow the major applications to be developed although these applications may need revision as further protocol possibilities develop.
The use of Ethernet into the IBM has yet to be proved. In addition the performance of such an IBM interface is not known.
No software is known to exist for the IBM or VAX which follows ISO standards and provides the types of facilities required. All known VAX software provides proprietary protocols such as DECNET and the provision of such protocols on the IBM is unlikely and, moreover, unwelcome to those attempting to promote community standards. Thus software costs are likely to be high in terms of manpower.
In the light of the uncertainties a phased approach to development is proposed giving a number of break points before a final commitment to proceed. When a final commitment is made it is envisaged that software staff will be required on the VAX and IBM for a protracted period and should help to develop the use of Ethernet at Rutherford beyond the requirements of HEP and into other areas where such a local area network is desirable.
Commission Ethernet interface on Central Computing Division VAX and perform loop performance tests. The following results have already been obtained using the HEP Division's existing Ethernet:-
Commission Ethernet interface on IBM and perform loop performance tests.
Connect both machines to the experimental Ethernet now being installed in Computing Division and perform compatibility and performance tests. The performance tests will be to determine maximum data rates and be similar to the VAX to VAX tests described above.
The VAX work will take one man month of staff effort assuming a good knowledge of Ethernet (for example using experts from HEP Division) or two man months of less skilled staff (for example one of the Computing Divisions VAX experts not familiar with Ethernet). The IBM work will take 3 man months of skilled staff. The IBM will take more effort because of the complex front end nature of the hardware.
At the end of this period the feasibility of using Ethernet and its possible performance will be known and a decision can be taken whether to proceed.
Implementation of the LLC1 protocol and the connection less network layer. This work may also include the connection less transport layer depending on the protocol experts decisions.
The VAX implementation may well come from CERN but since the work there will be in a development stage Rutherford will have to have considerable knowledge of the product. It is therefore expected that 6 months of effort will be required if a CERN product becomes available and one year otherwise.
The IBM implementation will probably have to be developed at a cost of one man year. This situation may change if CERN become interested in an IBM interface.
The initial applications are expected to operate over the connection less transport layer. Some design work is required for each application and so exact manpower requirements are not possible to determine. It may well be that some standards in this area become apparent before work has to start.
It is expected that other protocols will be required such as the connection variety. These should be available from manufacturers as the community is intent on pressuring manufacturers for them. Rutherford should add to this pressure. None the less some skilled effort will be needed in mounting and exploiting these products especially in the early days. It may well be that one of the manufacturers may be reluctant to provide products and that more effort may be needed.
(a) VAX UNIBUS ETHERNET controller $3500 (b) AUSCOM Interface to IBM $17000 (c) QBUS ETHERNET controller for AUSCOM $2500 (d) IBM channel $? (e) ETHERNET cable $5000
All equipment other than the cable is available on loan from computing division for Phase 1. Subsequent payment for all or part of the equipment by HEP must be determined.
Phase1:- Installation of cable and other equipment 1 man week VAX software and evaluation 1 or 2 man months IBM software and evaluation 3 man months Phase2:- VAX software 6 to 12 man months IBM software 12 man months Phase3:- To be determined. Phase4:- To be determined.
Which division pays for the staff has yet to be determined. Since and if the project followed standards and is therefore of general utility computing division may be prepared to meet some costs as part of the development of a networking infrastructure.
The proposal is far reaching in that it is designed to provided far more than is strictly needed for HEP. This is seen as desirable as other uses for such a network are highly likely to develop. Indeed it may be that the Rutherford local area network strategy may become based on Ethernet. It is still far to early to take such a decision with unknowns such as the fate of the Cambridge Ring and the popularity of the Token Ring unknown.
(1) Interim Report of the Datagram Definition Group. C Piney. CERN. 29 Feb 84.
Chris made it clear that BT would not allow a service without suitable authority from DTI.
Given authority it would be possible to provide a service pending a resolution of the tariff issue although I felt that he would be reluctant to do so. Most of the meeting was taken up with a discussion on the tariff issue.
Chris said that the meeting of CEPT at Auchen was inconclusive and he gave me the impression that the PTTs where not as happy with the proposition of levying a zero volume charge as the EARN Board of Directors had been led to believe. He hopes that the matter can be resolved at a CEPT meeting in October.
The PTTs has been comparing EARN with the networks run by CITA and SWIFT. They had had problems with the licenses with these networks which had turned out to be embarrassingly advantageous to the airlines and banks and there was an unwillingness to enter into another 'problem agreement'. Thus the interest in allowing EARN for a restricted period and to maintain the option of levying a volume charge.
BT had a further problem, which may not apply to other PTTs, in that by their charter they are not allowed to treat one group of users preferentially. Thus if the academic community were allowed to use EARN then SWIFT and CITA could claim to be treated similarly. Discussions with Chris, I think, convinced him that there were valid reasons for the network activities of EARN to be treated differently since EARN was non commercial. If EARN were to be treated like SITA and SWIFT then we would be at liberty to carry commercial traffic. However BT do have a point that EARN will loose BT revenue in that traffic now going over PSS will be diverted to EARN. Chris volunteered the point that the PTT's X25 networks were not geared to bulk traffic from a tariff or performance point of view. Thus it could be argued that EARN was providing a service which could not currently be provided by the PTTs. He thought that the fast circuit switched networks would be better suited to bulk traffic but international links using that technology are a long way off.
We discussed the form of the tariff. Chris, I believe, genuinely wants to give us a reasonable tariff which is not much above the leased line cost. He, contrary to the feelings of the BOD, does not think that the PTTs will allow a zero volume charge. He was interested in basing the charges on 'duration' rather than 'volume' basis since it was bulk traffic and bulk traffic is better suited to a circuit switched network. If we consider that a dial up European charge is 35p/min from the UK and 58p/min to the UK from Switzerland and further we assume that the line will be utilized for 4 hours a day (more or less the figures from the CERN link) then the charge would turn out to be 17,400 pounds a year to the Swiss and 10,000 pounds a year to the UK end which is about the price we could afford. It is an interesting idea and in my view if it solves the problem then fine but I do not really see what the advantage is.
I came away with the feeling that I would dearly like to see exactly what happened at the CEPT meeting since the stories seem to be all different.
John has also been dealing with the JANET license and although I was at pains to impress on him that the two networks should not be allowed to prejudice each other he believed that the same rules may apply.
The rule that applies is that since all the EARN equipment in the UK is at Rutherford and is owned and operated by them, and as Rutherford is part of DES then DTI cannot issue a license to operated EARN since it is effect government issuing a license to itself which is not sensible. Thus Rutherford needs to obtain permission from the DES secretary of state who would no doubt consult with DTI. The acid test as to whether Rutherford can come under these rules is if Rutherford is taken to court on some matter then who would defend the action- if DES then the rules apply - if someone else then the rules do not apply.
From a practical point of view I now have to find out if the rules apply and then get someone in authority to write a letter to John Taylor seeking the authorization and proving the Rutherford position. Easy when you know how!
We went on to discuss the finer points of the problem. Taylor was sensitive to the policy that EARN should have no privileged position unless such a position could be justified. Like BT he was worried that CITA and SWIFT may claim that EARN was being treated advantageously. He was worried about traffic originating outside JANET could use EARN and that this traffic could currently be revenue earning for BT.
I believe that we are moving closer to getting permission to set up EARN but I am not optimistic that this can be achieved in the near future.
Dear Mr. Taylor,
Thank you for your help and advice at our meeting on Tuesday.
I have now investigated the legal status of Rutherford and unfortunately it does not look possible to treat it as a government agent. Apparently Rutherford Laboratory is a 'place of business' of the Science and Engineering Research Council (SERC). SERC is apparently a 'Body Corporate' set up by 'royal charter' and as such is able to enter into contracts and is liable for the consequences of such contracts. This status is confirmed by SERCs requirement to obtain licenses for various activities associated with its use of radio. The requirement for radio licenses would seem, on the face of it, to be very similar to the need for a license to operate EARN in that both allow us to overcome legal obstacles to our activities.
In the light of the SERC status it seems that DTI is in a position to provide a license for EARN.
I hope that Chris Hobson of BT has passed you, or Mike Wright, sufficient information on which to base a possible license. I can see that there may still be a few problems. For example, EARN is not a legal body and thus a license will have to refer to the services to be allowed rather than the name of the network. May I suggest a few sentences which would make me happy:-
SERC is permitted to install leased telephone line connections between computing equipment installed at the Rutherford Appleton Laboratory and equipment installed at CERN, University College Dublin and possibly other European non UK academic sites for the purposes of data communications under the following conditions.
I hope that these words give an accurate picture of what we are attempting to achieve as well as making the use of the lines sufficiently different from other networks, such as CITA and SWIFT, to prevent embarrassing comparisons. Of course, I shall be happy to discuss any other conditions you think appropriate.
I hope this gets us a bit further along the road.
Yours sincerely
Paul Bryant.
A few things came up at NDM.
We talked about the telephone exchange equipment. We now need user documentation and I think that this should be produced in conjunction with Angus Goldfinch as he has a suitable series to put it in. It should contain a warning of the date by which the present service will be withdrawn. This date should be agreed between you and Peter Gill. I guess early in the new year if all goes well. (My recent experience is that it does not fail although it seems to take a very variable time to reply). It may be a good idea to see if David Rawlinson can provide figures of usage. We ought to first get a few more people in the Division to use it. We shall also need some documentation for Telecoms so that we can get them to look after it. Can we get them or can you sort our what sort of maintenance agreements we can come to with the suppliers- or someone. I guess it may be best to have a spare or two and just send the faulty one back by post.
We spoke about the Ethernet and it is agreed that for the time being you should manage the experimental Ethernet. We want to run it in a very informal way and management really only means that you should be kept informed of what is going on and help them as necessary. You should also keep an eye on the tools and connectors and so on which will be kept in the work shop. Do you know if the ethernet cable is still under the floor near RLGK? If it is still there we could use the terminators and possibly other bits and pieces to avoid spending money. I also have a bit or two in my private collection which may be useful. We are going to start an Ethernet Chat Shop which will be an informal meeting held now and again to talk about how the various projects are going, I guess you should be involved in it.
I went to an SNA seminar today when IBM talked about their new cable standard. We MUST take this into consideration in planning the site backbone network. IBM basically have two cables. A screened twin twisted pair and a twin fibre optic. The Twin twisted pair comes in various standards for inside use, outside use, flexible use and fireproof use. The cable can be used for 2780 (instead of coaxial), for the token ring, instead of twin ax. In the case of it replacing coaxial a passive adapter is needed. IBM has standardized a connector to go with it. IBM do not supply the cable or connectors but they can be had from BICC. I attach useful documentation. It is urgent that we evaluate this technology to see if it can provide a more flexible wiring standard for use, whether it is cheaper and so on. We need to know how it fits in with your site wiring plan. I think you ought to get some prices from BICC of the cable and connectors and well as a few samples for us to have a look at.
Paul.
Many of you may have heard of the European Academic Research Network (EARN) and wondered what is, what use it is, can I use it and why does it seem so unavailable.
First some background. There is a network in the States called BITNET which is formed from interconnecting IBM computers with leased lines. The protocol used is the IBM RSCS protocol and, in fact, any computers offering this protocol can join the network. The network relies on the fact that the RSCS protocols can be used for 'staging' files from machine to machine until a destination is reached. Thus the network provides a file transfer and mail service. The network is very popular in the States and some 200 or more machines are connected. IBM's corporate network, VNET, is based on the same idea.
In late 1983 IBM generously offered to finance a European network based on the same idea and christened it EARN. They offered to pay for the lines, modems and some other hardware for a period of four or five years. It had been hoped to get the network operational by the new year. The network is run by a Board of Directors with an academic representative from each country.
The plan is to link various key European sites in a backbone network and for each country to develop a network around their national site. In the case of the UK the site is Rutherford Laboratory. In the UK it has been decided that Rutherford should be the only EARN site but that a gateway should be developed between EARN and JANET to allow file transfer and MAIL traffic. The EARN network is connected to BITNET via a line between Rome and City University New York.
Unfortunately the network involves 'third part switching' of data which is illegal in most European countries and thus special permission has had be sought from the PTTs. Eventually CEPT (the European PTT advisory body) considered EARN and while they are likely to agree to allowing it wish to impose a few conditions. These conditions require the network to migrate to the use of ISO protocols and the use of public networks within the next four years. The EARN Board of Directors are sympathetic to these requirements. A further complication is that the PTTs are unsure of what tariff to apply and are thinking in terms of a volume related one. This is not the end of the negotiations since each of the PTTs now has to be persuaded to issue a license. You must remember that CEPT is only an advisory body. In the UK the Board of Directors member is busy encouraging DTI to issue a license. He is also busy attempting to encourage BT to allow connections and at a low tariff.
Currently the network is half built. The Rome to New York, Madrid and Heifa lines are working. The CERN to Rome, Paris, Dusseldorf and Rutherford are almost in as is the Dublin to Rutherford one. Unfortunately the lines to Rutherford cannot be used without BT and DTI approval. In Germany some 50 machines are already connected. In other countries less progress has been made and far fewer connections are expected. As well as the countries already mentioned connections are expected to Belgium, Sweden, Norway, Switzerland and Finland.
By now you will have realized why there has been a long delay and what sort of service you can expect when it all come to fruition. The UK EARN Board of Directors member is Paul Bryant who is happy to provide further information.
Dave Duce has passed Paul Borrill's application to me for comment.
Mr. Borrill has many credits to his name:-
The officers and members of IEEE bodies are supported by their companies as the work they do is for the common good, brings glory to the company, gives the company some influence and gives them early access to developments. Unfortunately Mr. Borrill does not belong to the sort of organization that can afford this support and he therefore has to seek support from elsewhere. DTI and SERC are two obvious funding bodies and the DTI funds appear to have dried up.
In my view it is vital for Mr. Borrill to be funded as:-
As I am on the IEE Computing Standards Committee I have been very aware of his work. I have been impressed by what he has achieved and optimistic that he can achieve much more in his position of Secretary which will be to the UK benefit. He has outstanding technical abilities as well as being an astute politician.
In reading the 'case for support' I find it accurate from by knowledge of Borrill's activities. The work undertaken under his Advanced Studentship, which has been closely associated with P896, has been very successful and supports his request as well as giving confidence that his request for further support will be money well spent.
I support Mr. Borrill's application strongly. It would be very disappointing if funds could not be found.
Present:- P Bryant Rutherford
P Girard Rutherford
C Setford IBM
P Hilton IBM
E Bodger IBM
There were no corrections.
M3.3 Permission to use the lines is still being sought. Meetings have recently been held with BT and DTI. BT will not allow the lines to be used without a license. They are willing for the lines to be used pending a resolution of the tariff issue. BT intends to put forward a tariff structure based on the duration of use of the line rather than the amount of data. Initial calculation show that the charge will be a bit more than the leased line charge. A meeting with DTI encouraged the view that Rutherford did not require a license on the basis that it was an 'agent of the Crown'. This has been shown to be untrue and a letter making this point and requesting a license has been sent to DTI. The level of contact with both BT and DTI is satisfactory but as yet unrewarding.
M3.4 A draft for the Network News has been produced and is appended. It will probably be compressed and amended before publication and in the light of any developments.
M3.4 Herb Budd has been approached on the production of a press release and has indicated that one will be forthcoming when CEPT and PTT problems have been resolved.
M3.5 Not done.
M3.7 It now seems unlikely that permission for a direct connection between Hursley and JANET will be obtained. Options for a connection via PSS are considered below.
M3.8 IBM is now expected to support staff effort at Rutherford and Rutherford awaits a letter.
Action: P Bryant
M3.9 The modems have been shipped to Rutherford.
M3.9 A multiplexer has been requested and will be supplied. Rutherford operations group will be asked to reserve floor space.
Action: P Bryant
Oxford University Press have a project supported by IBM which is associated with the University. It will have an IBM 43xx located at Oxford. Access to EARN is required.
A direct connection between the IBM 43xx and the Rutherford EARN node will probably not be permitted by the Joint Network Team in the interests of only allowing a single academic network with a single set of high level protocols. However, a connection can be demanded to JANET, assuming the project has suitable academic status, which may be funded to some extent from JANET resources. 'Coloured Book' communications protocols will have to be run and the code may be obtained from Rutherford. This will allow access to EARN and hopefully Winchester although it will not be quite as convenient as a direct connection.
Eric Bodger should write to Peter Linnington of the Joint Network Team at Rutherford requesting a connection and outlining the academic nature of the project.
P Bryant will discuss the connection with Peter Linnington.
Action: P Bryant
The Imperial College IBM 43xx will probably not be permitted a direct connection to Rutherford or EARN but can demand a connection to JANET. It was noted that there are already several such machines connected to JANET. In addition 3270 connections may be made across JANET between such machines with reasonable responses.
In view of the political difficulties with a direct connection between Hursley and Rutherford a connection via PSS is agreed to be the best alternative.
Four alternatives were considered.
The use of SNA or recently announced ISO proposals were rejected on the grounds that it appeared as though considerable research and development would be required.
The use of the Salford package was rejected on the grounds that is was a new and untried product and thus carried a high risk. It was felt that the project did not want to be a guinea pig for the product when their were safer alternatives.
Use of RSCS protocols over X25. In this option Rutherford would provide an RSCS to Blue Book gateway. The option would require some work to mount RSCS over X25. Although opinions differed as to the magnitude of the task it was certainly non trivial. The option had the disadvantage that permission to run RSCS over JANET would have to be sought and although this was not impossible the issue of allowing such activities was sensitive as it could be seen as setting a precedent. The option was rejected as there was a easier one.
Mount Rutherford Coloured Book protocols at Hursley. This option had the advantage that no software need be developed and the option could be tested at minimum cost and in a short time scale. This option was adopted.
It was agreed that a 37XX multiplexer would be used but that the Compro software would not. It was considered that the cost of the software was high and that there may be political problems in IBM using such products. There is also a migration problem if Compro software is to be used on newer IBM multiplexers. The use of this software can only be justified if a high throughput is required. It was agreed that the best option was to use an Edinburgh Box to convert Bi Sync to HDLC. This Box can be obtained from P Bryant at Rutherford Laboratory and will cost about 3000 pounds ex VAT. It has a maximum capacity of 9.6K.
The proposed program of work is:-
Action: Most
7 January 1985.
M3.3 Ask permission to use lines P Bryant
M3.5 Investigate use of Rutherford for mail
pending the gateway code P Bryant
M4.2 Expect letter from IBM about staff support P Bryant
M4.3 Discuss Oxford connection with JNT P Bryant
M5.3 Progress Hursley Rutherford connection Most
I went to a meeting at Hursley. Minutes appended.
Interesting points. We decided that the connection from Hursley to Rutherford should be via PSS thus sanitizing the data. You will recall that at the last meeting Les Clyne said to he thought a direct connection to Rutherford would not be allowed by JANET and although he was asked to confirm this I doubt if the situation has changed. They intend, subject to management agreement to put up Peter Girard's code and use an Edinburgh 'box'. This seems to me to be good as it gets a bit of our thinking into IBM and gets the connectivity we want.
IBM have now had a private meeting in Madrid and have decided that a coloured book connection is not a good idea (I have no reasons yet) and instead intend to have a leased line connection from Hursley to somewhere else in Europe to get over the problem. On thinking about the idea I wonder what the difference is between a direct Hursley-RL connection and an indirect Hursley- RL connection is from the point of view of JANET. This seems to be a ridiculous situation and a waste of money. Have you any further thoughts?
I spoke to Mr. E O Bodger who is an IBM employee and will be going to Oxford to work on some massive project at the OUP. I guess it is the computerization of the dictionary. IBM have donated a 43xx computer. He want a connection to EARN. He believes that he is academic not commercial as the projects has strong links with the University. I told him that if he was commercial he would not be allowed to connect to EARN or JANET. I also told him that if he was academic then there would be opposition to a direct RSCS connection to Rutherford but that he could demand a connection to JANET and that JANET (or Oxford University LAN whatever it is) would respond.
He will be writing to you along these lines and proving that he is academic.
It seems that there may be a few more requests coming up and I hope my treatment of them is to your liking.
I attach a draft contribution to Network news. It will need revision to bring it up to date just before publication and may well need cutting down a bit.
I also attach a report of my meetings with BT and DTI. I have now established that we do need the license and have told John Taylor so. At least I think I am now at the right level in DTI and John seems a reasonable guy.
I attended the ECFA Subgroup 5 (links and networks) meeting at CERN.
At the meeting the 4 LEP experiments outlined their communications plans. They all use large numbers of VAX machines spread round Europe.
ALEPH hope to set up a DECNET network round all their machines in part over X25- UK included. They claim that they can use JANET for this as Mike Wells and yourself (?) stated at some meeting that JANET was not concerned with the protocols running over X25. They also claim the precedent of Starlink using DECNET over X25. They have apparently been running some tests between their machine at CERN and a Starlink machine.
The other experiments have resisted the temptation of DECNET over JANET on account of JNT's desires. I detect some mild pressure in these groups, particularly the Italians, to have total DECNET systems and, of course, there are many attractions.
I bring this to your attention as I believe that allowing this use of JANET can be seen as a precedent whereas the Starlink use is a historic remnant. If you wished to resist SNA this would weaken your case.
The ECFA SG5 (Links and networks) meeting was well attended and it seems to be a rule that CERN meetings are easier for most people to get to. It was unusual in that as well as the normal working meeting there was also an open meeting. The group is now quite successful as it has started off a number of useful activities in the communications field. The ESONI meeting was aimed at seeing how much support there was for ESONI to support a local area network initiative. Earlier meetings had encouraged the idea that local area networks would be important in laboratories and therefore some standards should be supported. The meeting confirmed this view and some further plans were sorted out.
The INFN network meeting at Barrie will have an international day at which James Hutton will speak.
There was more speculation about the 'European Science Foundation' meeting which will be held at Coseners House.
We had some discussion on PTT tariffs without much outcome. It is clear that the packet switched tariffs, particularly on international calls, are too high and that if they were reasonable ECFA would prefer to back their use. It is clear that if low night tariffs and/or bulk discounts were introduced the services would be more attractive.
The GIFT machine has now been delivered. The GIFT DECNET code is very well advanced. The Blue Book Q end code is almost ready but the P code is more difficult than expected. A service could probable be provided by the end of the year. The UNINET code is not started yet. There was discussion as to whether TCP/IP should be put up but this will be considered later.
Although the interactive recommendations were complete and had been the basis of the Zander work there was some surprise that the final document had been produced without Ulf's consent!
Steve Kille is going to CERN for 4 weeks to look at mail. CERN are putting in a proposal to ESPRIT in conjunction with Kirstein, ICL, DFN, INRIA, Siemens and Bull to do something on distributed mail name servers. Sweden is interested in how X400 interacts with GILT.
Sweden will be joining EARN.
Swedish packet tariffs have been lowered for duration and raised for volume.
SUNET, which is up and running, is having no active development. They are trying for funds to develop FTP, X400, information services and a proper organization.
Imperial College reported that DECNET services were 30% slower than colored books on a VAX.
The Dutch internal packet tariff had been reduced by 40%.
Belgium seemed very happy with their yellow book products. They are about to put them on a DEC 10.
Brussels University were just going out to tender for a new machine.
Paul Van Binst reported that ESPRIT were disappointed that GIFT had not been put forward as a project. The meeting felt that it was really too advanced and that it was already suitably funded.
Oxford are very happy with their Cambridge Ring based on CAMTEC equipment.
DELPHI have their VAX at CERN connected to JANET but this will be moved to the CERN X25 network soon. It will also use DECNET over the Ethernet. In all they expect to have 10 VAXes using DECNET, coloured books and UNINET. 40 institutes are involved.
In Denmark CENRENET only provided X25 terminal traffic but there is a lot of international traffic.
In CERN 4 machines are now connected to the X25 network and there are expected to be 15 by the end of the year.
There are expected to be 10 Ethernets by the end of the year and these are to be connected to CERNET using the FRIGIT boxes (these are 68000 VME machines). TCP/IP is likely to be popular of the Ethers. I was offered the CERN Cambridge Ring equipment at a knock down price.
Manchester were keen to see full screen services develop and to see higher speed lines to CERN and DESY.
In Italy there is still little or no X25. All communication are on leased lines. The PISA VAX will have the JNET (not JANET) software to connect to EARN. They are front ending a UNIVAC IN Milan with a VAX to provide DECNET on the UNIVAC. All the Italian IBMs will be connected to EARN. They are keen to see 3270 services over DECNET to IBM VM/CMS machines (I wonder how!).
The Aleph experiment involves 6 countries and they want a distributed system over all the VAX machines involved. They see the only solution as DECNET and intend to run it over JANET and cite STARLINK as a precedent. They are worries over the regulatory aspects of the plan as well as the bandwidths of the connections.
The Dephi experiment involves 14 countries and 34 institutes. They are most interested in an integrated mail system and see GIFT as important in helping this aim. The Frigit development is important for interconnecting their machines on the CERN site. The experiment will use 10 machines on the site and these will be interconnected with ethernet. They are anxious for suitable protocols for use in the experiments to be developed.
Opal want to use networks for software distribution, software maintenance and for shipping small files. They also want to see full screen terminal services. About 30 institutes are involved. They do not want to set up their own network.
The L2 experiment want terminal services, mail and file transfer now at a rate of 1 Mbyte a day. They will need 100 Mbyte a day eventually which may rise to 10 Gbyte.
There was quite a bit of discussion on the PTT X25 services and it was decided to write to the PTTs asking for their plans for CCITT 84.
David Lord presented the Appollo project. This is a send only satellite system. The user requests a file over terrestrial links and at some time the file ts broadcast. If there are any failures these are reported over terrestrial links to obtain a retransmission. This system keeps the cost of the receivers small but is really only suitable for obtaining large amounts of data that is not urgent. There will be a pilot service in 1986. The system will have 4 channels at 1.92 Mbyte each. DG XIII of the EEC is involved in the project which is on the ECS2 satellite. Tariffs will be very low.
Jack Prevose talked about the use of the Telcom 1 satellite which provides an ISDN style of service. Cheap services can be obtained if booked well in advance or expensive ones on demand. A variety of speed from 2.4K to 1920K can be provided. Services will be available in 1985 and there will probably be a service between Saclay and CERN. There was quite a bit of discussion on how a network should be connected to the service which was mainly about whether files should be staged or whether end to end connections should be established.
This was chaired by Ted Owen and was well attended by 80 or so people. The aim was to decide what sort of support an initiative in this area would get. The answer seem to be a lot but that most of it was sleeping support. In the end the same old gang got together to plan the next phase. It looks to be a long haul. It seems well accepted that the 'Data Gram Definition Groups' idea to follow ISO up to at least the connection less network layer is accepted. Above that a lot of work is needed. There was a lot of discussion about the needs of applications which was a bit sterile as there were few hard facts. Most applications were described in qualitative terms and some metric for applications was needed to analyses the needs.
It seems certain that something like remote procedure call mechanisms will be needed but there seemed little expertise in the audience on the subject. Apparently there are some interesting CCITT protocols in the area called the Q700 series that I had never heard of.
The next meeting will be in Amsterdam.
It looks like the 'working set' will be CERN, Rutherford, Daresbury and Amsterdam. The next stage will be to encourage a few pilot implementations like the VAX IBM one at Rutherford and to attempt to clarify the needs of applications with a view to eventually defining suitable high level protocols.
Dear Mrs. Stables,
Thank you for your letter. There are a number of topics which you should be aware of.
I hope you do not find this letter to pessimistic- in practice I am sure we can work something out. I will be very happy to help you further if you would like to write to me or 'phone me.
Yours sincerely
Paul Bryant.
As division head you unfortunately have the Rutherford Communications Committee to run. Cliff was going to do it and I have prepared the Terms of Reference for your agreement.
Cliff suggests that I send them to Peter Linnington and to Division Heads for nominations. Please advise.
CTIG has certainly got a new lease of life! The hot topics were the TELEX Network Adapter, X29 over session(?) and input to CCITT.
There was a strong and angry feeling that BT takes little account of CTIG. We are fighting hard to make life as uncomfortable as possible for them until they recognize us.
Further good entertainment was had about the bugs in the PAD. After asking us to put them in priority order they still refuse to give us a list. BT claimed that we were told when bugs were fixed and to a man the meeting said they had never seen such a document. They then claimed that bug fixes were publicised as updates to the TUG which brought more mirth as we could not understand what bug fixes were to do with a standards document and morover no one had ever seen such updates. BT also made the amazing claim that they only put right bugs that did not effect users and ones that did effect users had to wait a major release of software. In all 40 love to CTIG.
After a lot of discussion the faults in the TNA were agreed to be faults by BT. BT were pressed hard to say if they would put them right. The best we could do was to extract a promise to report back within a month. If BT did not then a very strong letter would be written. We are fed up to the back teeth over BT failing after 2 years to sort it all out and our patients is exhausted.
There was absolutely no progress on international traffic and it is now my job to contact BT marketing (one PETER GLADMAN on the fourth floor of Seal House least I forget).
BT have put forward proposal for parameter settings and new parameters for block mode working. These are just the job for things like the full screen CIFER and Viewdata. They look reasonable.
BT also talked about point to point working which turned out to be a protocol for terminal to PAD traffic over dial up to give a network(?) interface. It looks to be a direct competitor to the async stuff invented to take the place of async FTP. It was a case of the left hand not knowing what the right is doing. Ether the JNT sponsored Async stuff should be taken or the new one taken and the loser abandoned. BT seemed unaware of the JNT/DTI one.
Unanimous that this should not be done and it appears that DTI have dropped it. So it must go over network or session. We thought network was the place as this would cause least problems when going to CCITT 84. However the nigger in the woodpile is what happens on LANs. If X25 is used then X29 over network is sensible. If transport class 4 then X29 over transport is a better bet as this will ease gateway problems.
Funny how all problems seem to be dependant on something happening elsewhere.
My superbly written paper which advocates that all X29 parameters are made mandatory as well as their settings caused a 2 hour discussion. It also advocates making some illegal parameter settings legal, the outlawing of national parameters and putting the version of XXX in the protocol id.
It took a long time to get round to agreeing that we should differentiate between our strategy and tactics to implement the strategy. The points on making everything mandatory was agreed as well as the outlawing of national parameters. It was felt that it was unlikely that we could achieve our objectives and the tactic will be to break the proposal into a large number of sub proposals and get through CCITT as much as possible.
I felt I made some good mileage.
Dear David,
I attach a draft EARN migration paper. I would value your comments as to its form and content. If it seems OK I will circulate it and we should put it on the agenda for the next EARN BOD meeting. Alternatively we could put in some more work to explore some or all of the options further.
At the BOD meeting I would anticipate a rambling discussion ending up with some decision to set up working parties to explore each option. Each party would attempt to promote their option and would present a case at the next but one BOD meeting when we could hopefully fix on one, or at least a small number of options.
Please feel free to comment freely. However I think we should aim at a dead line for a document of the end of November so that it can be circulated well before the next meeting.
Yours sincerely
Paul Bryant.
Dear Dennis,
I attach a draft EARN migration paper. I would value your comments as to its form and content. I am also sending it to David Lord. If it seems OK I will circulate it and we should put it on the agenda for the next EARN BOD meeting. Alternatively we could put in some more work to explore some or all of the options further.
At the BOD meeting I would anticipate a rambling discussion ending up with some decision to set up working parties to explore each option. Each party would attempt to promote their option and would present a case at the next but one BOD meeting when we could hopefully fix on one, or at least a small number of options.
Please feel free to comment freely. However I think we should aim at a dead line for a document of the end of November so that it can be circulated well before the next meeting.
Yours sincerely
Paul Bryant.
At NDM we had a long discussion on the reported bad performance from terminals.
It was clear that there was considerable confusion over what the problem was and even more over how the problem should be tackled. It was unclear as to whether the problem was confined to the IBM, to full screen Cifers, to connections via PADs or whatever. In short I got the impression that there was a lot of groping about in the dark with not much idea of what was being sought.
In a subsequent discussion with Pete Blanshard and Pete Girard I have come the following conclusions which may be incorrect:-
It was predicted back in 1976 in the Rosenbrock Report, which set up the Interactive Computing Facility, that the use of interactive multiuser mini computers was but a step towards the single user desk top systems. This prophecy is now being fulfilled.
The IBM Personal Computer is just such a single user system with the distinct advantages of having a vast installed base, a vast amount of software of a high quality, and a large number of add on peripherals. Moreover, compatible products are made by many manufacturers. Seeing its popularity it was inevitable that PC would start to appear at Rutherford and that it would be sensible to support then centrally and try and decide on some development program.
The PC is developing fast and so any efforts along traditional lines of writing user manuals and setting up large software development programs was inappropriate. The scheme adopted was to provide assistance and advice to purchasers and to undertake the administrative chores of keeping track of what equipment and software had been obtained. In addition expertise in PC software and hardware was obtained by investigating a number of products and undertaking a few minor hardware and software developments.
The principle interest is to see how best the PCs can be integrated into the networking structure of the Laboratory. The first, and easiest scheme was to make the PC act as an IBM 3270 both by a coaxial connection and via asynchronous connections. This also gave a crude form of file transfer. The more exciting developments are to connect the PCs to a high speed local area network using methods which follow approved JNT standards. This work is still in the planning phase.
I now have had a reply from most of the divisions and we can now set up a meeting. We need to produce an agenda which is difficult.
Terms of reference (do we circulate Peters memo?).
Do we consider all site networking?
Do we consider relations with other bodies such as JANET, CERN and Swindon.
The requirements for a local area network.
Who wants it?
What capacity?
What time scale?
What technology?
Who pays?
Survey of existing LAN activities.
What sort of coordination is appropriate?
Should we demand that any particular protocols be used or not used?
What scope and support is there for further pilot projects?
Terminal access.
Should it be via PADs, PACX, 3270 or what?
Do we need to upgrade our X25 on site network to improve response?
Administrative network requirements.
In discussing the requirements of a local area network demanded by various applications it has become clear that they are expressed in subjective terms. It is therefore difficult to match the requirements with the technologies which are available. This paper attempts to define a set of measurements and properties which can be applied to the application and the technology.
Classes of measurement are defined. The divisions are rather arbitrary and may not be appropriate in all cases. They are:-
Rest missing
Yes- we are aware of the problems of connecting the Auscom to the IBM. During the early stages we will develop driver code which is certainly capable of stopping the IBM and would certainly only be testable in development times. It is to be hoped that this driver will become stable and reliable to give you confidence that code at a higher level cannot effect the IBM adversely. How we give you this confidence is an interesting point which I would value your opinion on.
Much of the higher level software can be developed without the interface but in the later stages they have to be brought together. In principle this should not effect the IBM as the driver will have been well tested but we will do the initial testing only in development periods to gain further confidence. Once we have this confidence then there in no reason why the high level code cannot be developed in normal running.
I hope this is satisfactory.
We have not yet decided on the cable to be used. We have looked at fibre optic equipment and we are also looking at a 'thin' ethernet cable which could offer some cost savings. I am rather reluctant to drag an ethernet coaxial cable if we could use a 'standard' fibre optic pair. This begs the question as to what is a 'standard' fibre optic. Peter Blanshard is looking into this.
Area meetings of JANET users are being held. We belong to the South West Region which does not include London.
The meeting was chaired by Morley Sage of Southampton and Elain Williamson of Cranfield is Deputy. The meeting sends two representatives to a national meeting. Both the regional and national meetings will be at 4 month intervals.
The meeting is expected to be the primary input for users and sites to influence JANET. It is not expected to deal with day to day problems and complaints but only ones of longer term interest.
There were 23 representatives plus representatives of the London Rutherford and Cambridge NOCs. All but one of the reps were from computer departments and only one was a user. In fact it is hoped that many more users will attend and ideally there should be a centre and user representative from each site.
The meeting had no agenda and was therefore rambling. There were no clear indications of what the agenda would be next time as no issues emerged.
Mike Wells started off with words of encouragement which were stating how he expected the meetings to operate and what the state of various aspects the network were.
In discussions a few points were aired. The problem of commercial traffic particularly for manufacturers wanting to use JANET to help maintain machines on sites. The band width of JANET was discussed. ARPA and gateway authorizations came up. The discussions were all very bland and nothing of interest came up.
All in all it was a very boring meeting. It is difficult to see how it is going to get momentum with only meeting at 4 month intervals. I suspect sites will find rather more direct ways of making their views known. Clearly the cycle time for complaints, cements or whatever is so long that the issues will be out of date by the time they get aired. Having the two level structure will also discourage the meeting being used for primary means of influencing things.
We have had 230 or so applications for the current advert. It appears that the publication 'Current Vacancies' rolls them in. The other favorite ones are 'The Guardian' and 'New Scientist'. 'Computer Weekly' only attracted 2 applicants and the 'Scotsman', which we didn't advertise in, only 1!
Short listing has been done and we now have 40 or so candidates.
We will be holding an 'Open Day' on Tuesday 27 November. In the morning they will be treated to a talk about SERC, Rutherford and all the things they ought to know about such as career structure, sports facilities, accommodation. Importantly they will ne told what Rutherford does and in particular what Computing and Informatics Divisions do. After lunch a couple of people from each group wanting staff will set up tables in the colloquium and the candidates will be able to talk about prospects in more detail. In addition, the groups will be able to pick out the people they are interested in and so help the subsequent board. There will be tours of the machine rooms arranged. After tea there will be a question and answer session. This will be followed by an invitation for the candidates to re-affirm their interest in a job at Rutherford. I have seen all the group leaders in computing division and a few in Informatics and hope the arrangements meet with your approval and that you will be able to provide a couple of people for the afternoon do.
Regretfully, I will not be available on this auspicious occasion. Jed Brown will give the morning talk. Jacky Hutchinson is doing much of the organization of the afternoon - getting name tags and notices set up. Dennis Williams will be organizing getting them here. Thanks be to them.
As usual, this has all been done in a dreadful hurry and is likely to be an organized shambles. I would invite you all to monitor this experiment closely to asses its success.
The list of candidates and their details are available from me.
Replies are still coming in and there are some 50 applications which have not been considered plus a large number from the short listing who are reasonable. These could well be fodder for the next trawl.
This has been festering for a long time. The basic problem is that Computing Division has been unwilling to make any statements until IBM financial support, JANET approval, DTI approval and BT approval were known. Thus JANET and Computing have been eyeing each other over the table with the 'after you Claude' syndrome.
The position is that IBM have given verbal statements that they will pay for the lines to CERN and Dublin and will give us modems and a 3725 multiplexer. In addition they will pay for up to 6 man months of effort to build the gateway. I have no written statement from IBM after 9 months of nudging but this does not worry me as IBM are not in the habit of breaking verbal agreements.
The DTI position is that they are trying to decide how to license EARN. They have been doing this since they took over regulatory affairs from BT. Indications are that I should get a reply in a week or so (believe that if you will). BT will not give us permission to use the lines until we have a license from DTI. However the lines are installed but not being paid for.
My position is that I attend the EARN Board of Directors meetings on behalf of the academic community and at Rutherford's expense. I have lined up Tony Burreston to do the gateway job which is not extensive.
The policy of having a single gateway into JANET is sound and one I insisted on when IBM first came to visit us. If we do not look after this gateway then the job will go elsewhere to another IBM site- ULCC say. At this late stage such a change would not ingratiate us with the community. I would add that the project has been aired at group leaders and at other times and the general feeling is that it should be supported.
I believe we should state that we will provide the gateway service subject to EARN being allowed to exist by DTI and BT and subject to continuing support by IBM and other European participants.
I am not so sure how we consult on the specification of the EARN JANET gateway. I am reluctant to call some meeting as I would not even know who to invite and what the aims would be. As far the technology is concerned, I intend to use a piece of code known as the 'Mailer' which is used in EARN and to adapt it to do the gateway function. We have a rather poor spec of this and Tony is looking at it. I would be happy to present what we are or have done at Networkshop which could be taken as consultation.
I would not like users to include the use of EARN in any plans until we have sorted out the BT DTI problems.
Welcome to the Rutherford Appleton Laboratory. We hope that you have a pleasant, instructive and profitable visit.
The purpose of this Open Day is to give job applicants who have been short listed an opportunity to meet the staff, see the equipment, see the working conditions and learn a little about the Science and Engineering Research Council. This should give you a good idea of whether you would like to work here.
At about 11.15, when we have all arrived, Jed Brown will talk to you about SERC, the Laboratory and the staff we are looking for. He will cover the conditions of work, social and sports facilities and many of the other aspects of life at Rutherford. Jed works in the Administrative Computing Group but since he has been with us for many years he will be able to answer your questions at the end of his talk.
At 12.15 we will take you for lunch when you will be able to try our excellent restaurant and talk informally to some of the staff.
At 1.30 we will split you up into groups. Some groups will tour the computing facilities while the others will be able to talk to members of various groups who are looking for staff. Regardless of what you put on on your application we will be very happy for you to investigate any of the vacancies.
At 3.00 it will be time for tea after which we will assemble for a further question and answer session.
We hope that you will enjoy your visit and we would like you to indicate on the form below whether, having seen us, you want to be considered for any of the vacancies.
The representatives from the Groups who will be available are:-
Peter Blanshard Micro processors, network hardware. Peter Girard IBM networking. Graham Robinson IBM PC developments and LANs.
Tim Pett and Dave Greenaway : System software for IBM, GEC and VAX Bob Taylor computers. Graphics.
Jed Brown IBM office automation system PROFS. Terry Dunwoody Transaction processing in particular CICS. Judy Lay Data bases and in particular library ones. Brian Read Large scale scientific data bases.
Jacky Hutchinson Documentation and user support.
Peter Kent Support of the SERC and Alvey program in IKBS
Mike Jane and Geoff Lambert Support of multi user GEC and PRIME machines.
Ken Robinson and Tony Williams: Evaluation of single user computers for use as high quality work stations.
Keith Fermor and Cyril Balderson : UNIX support and development
Liz Fielding: Research and coordination for Alvey program.
Jim Diserens, Brian Colyer and Chris Greenough: The Division has a number of vacancies for applications programmers.
The Computer Services and User Support & Marketing Groups of Computing Division are supplying guides for tours round the computer rooms. Keith Benn, Mick Reid and Garry Williams.
Thank you for agreeing to take part in the Recruitment Open Day on Tuesday 27 November.
I attach the document which will be given to the candidates. We are inviting the 40 short listed from the recent advert which attracted 230 applicants.
This open day is very much an experiment to see if we can improve our recruitment methods. The aims are to give candidates an opportunity to see us and if they do not like us to withdraw and save us the bother of boarding them. It gives us the chance to see the candidates in an informal atmosphere and to pick out people we are particularly interested in.
You might like to attend Jed's talk although I would not like to see the Colloquium packed out with Rutherford staff, it would not give a good impression. I would welcome you to take lunch with the candidates to make them feel at home. Regretfully this would be at your expense. I do need you in the Colloquium at about 1.30.
At 1.30 the room will be set up with a table or two for each group. Name tags will be supplied for you and the candidates. You should bring along any publicity material you think suitable.
PLEASE REMEMBER THAT THIS IS AN INFORMAL OCCASION AND WE SHOULD BE PROVIDING INFORMATION TO THE CANDIDATES AND NOT BOARDING THEM. None the less we should be trying to detect the winners.
I am sorry that this has been set up in such a rush. May I also apologise that I will not be available on the day (my excuse is that being chairman of the board I must preserve my impartiality, also I am abroad). Dennis Williams is organizing getting the people here and Jacky Hutchinson is organizing the labels and tables. As usual, Gill Keats will look after the 90% of problems we have forgotten about on the day.
Thank you for your cooperation.
I have had Tommy Thomas on the phone concerning a meeting you had with him about 3270 terminals for admin computing.
The gist of the conversation was that he wanted to develop a 'superior' product similar to the IBM PC product I have developed. He wants to mount it under MS DOS rather than PC DOS as they have a number of PCs using this system. He wants it to interface to TOP VIEW which is a PC product I am unfamiliar with.
His idea is to get one of his staff to do the work and he then has some idea of exploiting commercially.
I am not absolutely sure what he is up to. It is important that any work done follows some standard so that the two sites can share developments. I worry over the commercial exploitation as we might find ourselves paying others for what we have done!
A complicating issue is that IBM would like us to develop the stuff to provide colour facilities on the PC. This involves some work on the IBM by Peter Girard, further work by the HEP group at Edinburgh as they did the PAD on the IBM plus work on the IBM PC- me I guess. IBM are prepared to finance this work. The object is to allow some of their university projects to use JANET. The problem, as usual, is our lack of staff.
Can I urgently discuss with you our attitude to these things. My view is that we should do as much as possible here in order to build up our expertise. We have learned the hard way that 'passing developments to other sites' is a good way to loose control, loose moral and generally go down hill.
IBM are proposing to set up a site in Winchester for the express purpose of supporting the IBM computers which have been given to Universities. Thus they want a JANET connection to a 43xx there. This machine will not be connected to IBMs own network. I have agreed to explore the possibility of providing a JANET connection. The first problem is the political one- this connection does seem to be different to the ill fated Hursley one. The second question is protocols and I am led to understand that they would be prepared to run coloured book stuff and may request and be prepared to pay for help from RL. The third question is who pays and here again I am led to believe that IBM is not short of a bob of two for lines and modems.
A further suggestion is that IBM may like to locate the machine at some convenient site, such as Rutherford. So far I have not explored the financial arrangements that could be arrived at but before proceeding further is this of interest to RL? My view is that in the past we have seen universities getting all sorts of goodies from manufactures while we have stood on the side lines complete with halo and where did that get us?
Dear
Last time I wrote to you, which was back in April, I had hopes that we would have had a positive response from DTI to our High Speed Local Area Network proposal. Since then I have been pressing DTI for a reply. I have now had a verbal reply from Adrian Grille and they have decided not to consider the project for funding. In fact, Adrian tells me that no projects have been funded for a very long time and are unlikely to be funded in the future. I must apologies for the length of time that it has taken to obtain this disappointing news.
I have discussed the project with Derek Barber of Alvey and Nick Newman of Esprit. I have had no response from Alvey and would not really expect one since they appear to be embarking on other network projects. Esprit look like being a more attractive source of support and it seems that a proposal in early 1985 for the next round of proposals would stand a reasonable chance. The rules of ESPRIT are that the participants in a project must come from at least two community countries.
If you think it is worth proceeding then I am happy to do what I can to further a proposal through ESPRIT. I would value your guidance.
Since we produced the proposal the FDDI standard has been progressing and, in fact, a number of American manufacturers appear to be intending to produce kit. Thus if a proposal is to be put to ESPRIT then the original paper will need updating. It will also be necessary to find a non UK participant.
This letter is really aimed at the two companies who showed continuing interest after my letter of April but I thought the rest of the group would like to know what progress had been made.
Yours sincerely
Paul Bryant.
DG came and showed us their PC. It certainly is portable in that it is light and can fit in a brief case. It also has a low power consumption which allows it to run for 8 hours on its batteries (a point I did not test). The screen is not all that readable and you have to get the viewing angle right to see it clearly. It is IBM compatible with 25 lines of 80 characters. It cannot take IBM plug in boards but it does have two communications ports. It does not have a Sintronics port for a printer but the second serial port can be used for that purpose. We did not test that. The demonstration was impressive but did show up a number of problems which must be recorded:-
The effect of the incompatibilities depends on what you want to do. We ran Display Writer, Word Star, Personnel editor with no difficulty. I suspect that most products will also run easily.
Communications programs would not run and in particular the 3270 emulator. Items 2,3 and 4 caused problems. Problem 4 was easily overcome and was due to a store clash problem between PC and MS-DOS.
Problem 3 was overcome with a few lines of code, or at least we saw how to overcome it. Data General have been very silly in not making the keyboard identical to a PC one.
Problem 2 is more serious. They use a different coms chip to reduce power consumption (the PC one is bi-polar and this one is in CMOS). Thus any code which directly drives the chip has to be rewritten. This is not a difficult job. We estimate about a month of effort to sort it out. If this work is to be done then I am unable to do the work with out financial support.
Problem 1 is a problem if you want to transfer data and programs directly from an IBM PC to the DG one. I like the small discs better however if you do get a machine I would advise that you get a 5 1/4 disc with it for transfer purposes although you would not lug it round with you.
Another portable is being produced by Texas and this should be available in Jan. It is a pound heavier than the DG. Its compatibility rating is unknown but I hope to have a look at it when I can get my hands on it.
If you want to get a DG one then we would be prepared to undertake to modify the communications software for 3270 on the condition that a 5 1/4 disc drive is obtained and that finance for the effort if provided.
Although progress has been very slow there has been some progress. DTI has been heavily overworked with the privatization of BT. This has come at an awkward time for EARN. There seems to be no arguments of principle but in the early days of a privatized BT both DTI and BT are unwilling to adopt a relaxed attitude to 'non standard' services.
File transfer facilities between EARN and JANET will be available as soon as permission is obtained.
Mail gateway facilities will take longer to provide. Documentation on the Mailer suggests that the gateway can be provided by 'standard' additions to the product and will not require code changes. It has proved difficult to obtain a copy of the Mailer to undertake further work.
Country |Gateways|Active|Planned|Backbone links |Internat|Volumn Charge|
|planned |Nodes |Nodes |Active |Planned|licence |Local |Inter.|
| | | | | | | | |
Austria |- |0 |? |- |D |? |? |? |
A | | | | | | | | |
Belgium |- |0 |10 |- |F |No |Ex |Ex |
B | | | | | | | | |
CERN |- |6 |? |D C |UK I F |Yes |N/A |Ex |
CERN | | | | | | | | |
Denmark |? |? |? |? |D IC |? |? |? |
DK | | | | | | | | |
Finland |FUNET |0 |1 |- |S |? |N/A |Ex |
FI | | | | | | | | |
France |- |7 |37 |- |S B |No |? |Ex |
F | | | | |CERN | | | |
Germany |DFN |61 |15-20 |CERN |A IR H |Yes |0 |Dial |
D | | | |GWU |DK | | |up |
Iceland |- |0 |1 |- |DK |No |N/A |? |
IC | | | | | | | | |
Ireland |HERNET |2 |0 |- |D UK |Yes |0 |0 |
IR | | | | | | | | |
Italy |ARPA |10 |2 |CUNY |CERN |No |0 |Ex |
I | | | |E L | | | | |
Israel |- |13 |5 |I |- |Yes |0 |0 |
L | | | | | | | | |
Netherl.|- |1 |18 |- |D |No |Not Ex|Not Ex|
H | | | | | | | | |
Norway |UNINET |0 |1 |- |S |No |N/A |? |
N | | | | | | | | |
Spain |- |5 |3 |I |- |Yes |0 |0 |
E | | | | | | | | |
Sweden |SUNET |4 |1 |- |F DK N |? |N/A |Ex |
S | | | | |FI | | | |
Switzer.|- |3 |7 |CERN |- |N/A |? |N/A |
C | | | | | | | | |
UK |JANET |0 |1 |- |CERN |No |N/A |Ex |
UK | | | | |IR | | | |
Ex=Expected, N/A=Not applicable, GWU=George Washington University
CUNY=City University New York
EARN got set up rather more formally. A proper association will be set up so that EARN BOD can get some funds from IBM (and others?). Dennis Jennings regretfully resigned as chairman as he will be looking after a research network in the States for a year. David Lord became the new chairman. There was a lot of discussion on the EARN charter which was re drafted. CEPT and the tariff problems came up as usual. There is going to be a technical meeting possibly in the UK.
EARN headed note paper appeared as well as an EARN poster. The poster (8000 copies printed) was rejected!
A second line to America has been installed. It goes from GSI to George Washington University. GSI is becoming the major node of EARN because of the way the German universities have embraced EARN.
Relations with Hubner of CEPT are very cordial although the tariff issues still have to be resolved.
Problems still remain with the UK, French, Italian and Belgium PTTs. The French and Italian problems appear to be the left hand of the PTT not knowing what the right hand is doing.
Now that CEPT is better aware of the EARNs objects they now seem happy to let it continue. none the less they are still insisting that EARN move to ISO protocols and the BOD accept this. CEPT wants the various PTTs to fix the tariffs. This is not a desirable state of affairs as the tariffs vary widely from country to country. Half the countries are only imposing the leased line tariff whist others (like the UK) are imposing a tariff similar to the PSS one. This is the next battle ground.
The changes were all to remove problems with connections to other networks, removal of any references to particular manufacturers or products, and removal of BOD powers to dictate to members on matters not concerned with EARN.
Usually an EARN member must allow other EARN nodes to connect to him. In the UK case this has to be interpreted as sites being allowed to connect to JANET.
A legal organization is being set up so that EARN can accept money from IBM to finance various activities such as meetings, publicity and travel. The association will be set up in France. The chairman is David Lord, secretary Ipollito (france) and treasurer Trumpy (Italy).
Herb Budd reported that IBM will be making a set of floating point array processors available at the Rome centre. IBM are setting up a scheme in France to allow students to produce chips. IBM will make the chips and the designers will be submitted via EARN. It seems that IBM is going to use EARN to provide services to Universities. The inability of IBM research centers to connect to JANET is likely to prove embarrassing in preventing UK users getting at IBM facilities.
It was agreed to hold an EARN technical meeting for the people running EARN nodes. I was asked to arrange this and I will try and get it at Coseners. It will be a two day event in January/ February for 20 or so people.
We were shown the EARN publicity material. The letter head was good but the poster was rejected at it looked like an advert for an IBM PC. It will be re done.
Unfortunately my 'migration to ISO' paper has not yet been agreed with David Lord. It looks like the Jan/Feb technical meeting will address the problem.
I am amazed at the progress that EARN has made in contrast to the time it took to get JANET going. Although the aims are less advanced the services are quite good. JANET could also learn a lesson on publicity. Posters, gifts and free services certainly make friends. The informality of EARN contrasts with the bureaucratic approach of JANET.
Dear ,
At the recent EARN Board of Directors Meeting I was asked to set up a technical meeting for the people looking after and developing EARN.
I have arranged this meeting for February 18 to 20. We will convene on the evening of Monday 18th and finish just after lunch on the 20th. This will hopefully make travel easy.
The meeting will be held at the Coseners House which is a small conference centre in Abingdon run by Rutherford Laboratory. The centre will accommodate between 20 and 30 people which should allow one or two experts from each country. I would like to keep the meeting relatively small as I believe a small meeting is more likely to make progress and allow us to work informally. It may be necessary to limit numbers in which case I propose that countries not yet connected will possibly be limited to one person.
I am pleased to report that IBM, via the EARN Association, will fund the meeting (?) as well as travel costs (?).
I would like you to nominate your representatives and let me know their travel arrangements so that I can arrange transport from the airport and send them any further literature.
Attached is the agenda for the meeting. I am most happy to add any additional items. I would ask that experts in any of the areas come prepared to present their ideas and written contributions would be most welcome.
I hope you will be able to support the meeting and look forward to being able to coordinate the technical progress of our network.
Yours sincerely
Paul Bryant.
February 18.
Arrive. A buffet supper will be available and there will be an
opportunity to get to know each other.
February 19.
9.00 Introductions - Review of state of EARN - Plans for the rest
of the meeting.
10.30 EARN information services and NETSERV.
EARN mail services and the MAILER - Should we adopt RFS 822?
14.00 Problems with RSCS, JES2, JES3, JNET and other products.
15.00 Gateways to other networks - what possibilities are their
for cooperation? What products need to be developed?
February 20.
9.00 Centres of excellence - It may be appropriate to designate
certain people or sites as 'experts' on particular products
and provide help to others.
10.00 Migration to ISO protocols. As a result of being allowed to
use leased lines EARN has agreed to migrate to the use of ISO
protocols. It will be a job of the technical group to advise
on how this should be achieved.
14.00 Depart.
I have just returned from the Board of Directors meeting of the European Academic Network. I am the Technical Coordinator of the Board.
We decided to set of a technical coordination meeting for the people who look after the network and I was asked to run it in the UK if possible.
The agenda will cover:- Problems with the RSCS, JES2 and JES3 products. Network products such as the mail system (MAILER) and the information centre code (NETSERV). Migration to ISO protocols.
The meeting will hopefully run from evening Feb 11 to Feb 13. IBM will probably meet the cost of the meeting in full or in part. The meeting will probably have 30 or so delegates from up to 17 European countries. I would like to use Coseners House for accommodation and for the meetings as long as the numbers do not rise to high. I will have to organize this event and will require some secretarial services particularly on the day. I will chair the proceedings. I would like Peter Girard and Tony Bureston to attend.
The benefit of the meeting will be for European academics and for cooperation between countries. There will undoubtedly be some expense to the lab if only my time.
My I request permission to run this meeting on behalf of EARN Board of Directors.
Dear Derek,
Thank you for your letter and your interest in the high speed LAN. There are a couple of points I would like to make that I could not really make in a letter which had a wide distribution.
Frankly, and not to put to fine point on it, I have been disgusted by DTI's treatment of the project. The group was led to believe that there was a fair chance of getting funding and took advice in submitting the document. We were led to understand that we would be able to 'negotiate' the exact form of the proposal and the document you saw was merely a draft to get DTI's reaction. In the event we had no response despite repeated 'phone calls from me. Eventually in August, if my memory serves me right, I was invited to DTI and led to believe that if Rutherford could provide project management at no cost to DTI then funding may be possible. This I did but still no response. Eventually I put a 'dead line' to DTI by which time I would assume that DTI would not fund. This ran out several times before I at last managed to get DTI to comment. Certainly this has been an upsetting experience.
You may recall that Dr. Manning initiated the idea which received warm support from the management of companies. The working meetings were not so successful as a lot of the participants 'came for the beer' and had no intention of participating. NPL and Harwell fell into this group who were invited against my advice. However we did eventually decide to support FDDI whereupon Acorn, for their commercial reasons, decided that they could not participate. In the event ICL and possibly GEC and STC were left in.
Personally I was only the honest broker in this work and was not intending to have Rutherford participate except in managing. Hence I doubt if I will participate in John Burren's work- running and developing networks for service use is my expertise.
I am sore at having wasted 3 or 4 months driving the work and feel that I have been treated by DTI in a most cavalier fashion. However I bear no grudges and put it all down as life's deep rich passing pageant.
Now to less unpleasant matters. I do hope that the Acorn work is successful as the need for a high speed backbone network joining slower speed ones is now apparent at CERN and I see a need at Rutherford in the future.
I believe that you must ensure that the technology goes to a standards body as soon as possible. In addition I strongly believe that it must relate to the other high speed standards, such as FDDI, in the same way as the 10Meg standards relate to each other. That is, the higher level protocols should be able to run over all the standards with the only differences being the quality of service offered. The problem with the Cambridge Ring was that it did not go to IEEE so that it could be developed side by side with the CSMA/CD and other standards. I fear that, although it has now gone their, it might be too late to make it popular as the CSMA/CD and Token Ring have a lot of momentum. I hope that this mistake is not repeated.
Anyway, I wish the work luck and if there is any help I can give it I will be pleased to assist.
Yours sincerely
Paul Bryant.
The meeting was a review of NNTF's activities to date. They asked the question as to whether NNTF should continue and decided that it should. I am a little doubtful as to whether it should. There are now far to many meetings on networks and many of the groups are stepping on each other toes or duplicating activities. It also seems that working groups come under fairly arbitrary mother organizations. Representation is also a problem in that some meetings are closed and others open. The means of influence are unclear. In particular the representation of the man in the street or user is absent or poor. In some cases work is being led by a clique.
CTIG is hard at work on CCITT 88, TELEX and the new BT PADs. Its main problem is getting information from BT and persuading BT to take its recommendations in some cases. An interesting item of work is the provision of a user friendly X28 interface to put to CCITT. They hope to have a look at VPT when it is complete but they do not consider themselves standards producers.
The X25 group has had 3 meetings and are interested in X25 harmonization particularly in the use of the facilities fields. They are addressing the question of migration to CCITT 84. They are not unfortunately looking into the problems of international working. An interesting work item is X32 which is X25 over dial up.
NSSG have been concentrating on addressing problems of X25 84. There still seems to be questions on the allocation of numbers. It seems OFTEL may manage the UK set. High level names and directory services are being studied.
ISDN is now embodied in the I series of standards which the group is studying. Apparently some enhancements are being introduced in 1986 when some stability is expected. Work is needed to provide the network service over ISDN.
CCITT had now agreed the 3 TELETEXT standards at Madrid, known as T101. The CCITT red book will be available at 434 pounds (must get a copy). There have been a lot of committee reorganization.
CEPT has also been reorganizing. There seems to be CEPT and CEN/CENELEC harmonization.
ECMA has reorganized and the old TC23/24/25 committees have been replaced by TC32 with 8 subcommittees. Active work includes message interchange, PABX, and VPT.
IEEE now looking at 1 Meg CSMA/CD LANs, broad band, and the metropolitan area network which looks like it will be on slotted ring principles. The token bus is still being supported by GM and PROWAY. The HiLi group is studying the protocols above LLC and X25 and ISDN seem the possibilities. Management protocols are being developed.
I had a long and useful discussion with P M O Garbutt of IBM who is in Telecommunications Strategy and Plans who is a very useful contact. He could be helpful in furthering our interest in IBM's ISO products.
Thank you for your memo.
Dave Toll, Graham Robinson and myself have been looking at the products you list and many others and have amassed a giant stack of waste paper to prove it. We have come to a number of conclusions:-
Our analysis of the products goes as follows:-
All the products fall into 4 groups. Bi sync, coax, full screen Cifer, and local area network.
The bi sync product imitates a 3270 on a bi sync connection. This requires the PC to have a bi sync connection to a 37xx, 1270 or similar. It can also be connected to other IBM and non IBM remote multiplexers such as 3274. Some manufacturers have produced look alike 3274 which is what things like ClusterNet and PC 3270 Cluster is. Some manufacturers, such as IRMA, have produced an interesting variant which connects further PCs via an async connection which cuts down the cost at the expense of not allowing a genuine Bi sync 3270 to connect to it. The scheme has the problem that the master PC must be kept switched on which could be awkward if the PCs are spread over several offices. The cost of the scheme must be set against a genuine IBM 3274. The speed of the terminal is governed by the communications line and is normally 9.6K. The connections require modems or modem eliminators on the lines. These connections cannot be used via PAXC. Thus leads to a rather inflexible set up as current wiring and equipment is not easily used. The performance of the terminals is a bit better than the full screen Cifer which is limited to 4.8K but experiments seem to suggest that this speed difference is not all that noticeable. Thus my belief is that the use of bi sync connections is more expensive than full screen Cifer and has no advantages (non withstanding the current memorex problem) and has several disadvantages. We have one IBM bi sync board which is less software which would cost another 500 pounds. The use of bi sync may be useful off site where there is already a bi sync link with genuine 3270 equipment but even here I would not recommend it if there are network connections nearby. I should add that on site I am rather against yet another means of connecting terminals - we have far too many already which strains telecoms resources.
IBM have a board for coaxial connections but unfortunately they only supply it for the 3270 PC and XT370 products. Thus we have to go to an alternative supplier for boards for the PC and PC/XT. As you are aware we have purchased IRMA boards. As you have noted, several other manufacturers are producing similar products at a similar price. The Idea 3278, Blue lynx, PC 3270 coax and Quad 3278 are just such products. The PC mags look at these products from time to time and have far better resources to evaluate products than I have and I tend to believe what they say. From the documents all the products seem very similar and I do not think I am justified in spending the best part of 1000 pounds and some scarce human resources on looking at each product. In addition further effort would be needed to integrate it into our MOS product. We would also have a further type of product to look after and I want to reduce the variety of equipment we have to reduce the costs of looking after it. Our feeling is that the IRMA board is as cheap as any and has a good reputation. It is much better than some of IBMs early products which we unfortunately have in the batch of XT 370 just delivered. (The problem is that the IBM board takes 2 DMA channels and this interfaces with the disc.)
I am sure I do not have to explain the full screen Cifer situation. It is a thousand pities that it has had the bad luck of the Memorex problem, the bad experience at ROE which fortunately is solved, and the slightly unsatisfactory Cifer implementation. Unless you know differently, the PC version looks good and reliable and gives a performance comparable to a PC on a bi sync link (again- not withstanding the Memorex problem). Certainly the scheme has the advantage of flexibility of being able to attach to almost anywhere in the network to obtain 3270 or glass teletype services. The most important failure is the lack of colour. Unfortunately this requires modifications to the IBM network code and to the IBM PC code but it is possible. Unfortunately Peter Girard is too overloaded to do the work just now but I am investigating whether I can get IBM to support the development which they require to allow 3270 facilities into some machines they have donated to universities.
All, as far as I can see, local area network products are based on Newcastle connection techniques. That is, a piece of code is introduced which intercepts all the system calls and can then possible direct them to other machines. A neat and very effective and cheap idea. It happens to be very easy to implement. Failings in PC DOS give some problems such as lack of file locking facilities. Newer versions of PC DOS overcome these problems. Most suppliers have put in 'added value' to their products in the form of office automation products. I would mention products such as Torus which are of this nature. These products have the almost universal problem of not connecting to anything else. In particular they do not provide 3270 or even glass teletype facilities across such a network. In principle it can of course be done but I would be unhappy undertaking such a task as I am very reluctant to produce such products when efforts should be going into promoting ISO. ISO might be on a long time scale but at least we can utilizes products from else where. Most of the products use proprietary network technology and this, unfortunately, includes IBMs. A couple of manufacturers do provide standard technology. Logica's product is based on the Cambridge Ring but certainly does not follow the upper layers of CR82 so this does not present a way of using the Ring for interconnection with other equipment without substantial developments. This I am reluctant to undertake on two counts. First, I am still smarting from my earlier Ring extravaganza and require much more assurances that the Ring is 'popular' and 'standard'. Second, I do not have the manpower. Torus is based on IEEE ethernet so is standard. In fact, the hardware is based on the 3 COM kit. This regretfully is a load of rubbish. The best ethernet product is the BICC one. I am getting a pair of these for two purposes. First it will monitor the IBM to HEP VAX. The second purpose is to build up ether net expertise. We are intending to experiment with a few of the connectionless ISO protocols and probably mount IBMs PC LAN code over it as a vehicle for experiment. This is a product compatible with PC DOS which provides the sort of facilities that the IBM PC network provides but, of course, on a 'standard' network which we hope to develop. Another point we have to consider is whether we need to develop clusters of IBM PCs just now. As I see it, most of the PCs are isolated and have no need to communicate. The only requirement I can detect just now is to connect my machines for transferring software. This is not a strong requirement and one which we would prefer to develop on 'standard' technology and hopefully and certainly eventually 'standard' protocols as authorized by JNT. Incidentally, the work with Ethernet has the possibility of having file server. With the connection to the IBM it may well be possible to make the IBM a file server to the PCs in such a way that the PC would see the IBM as its disc. This has clear attractions particularly if the files can be shared. Before you get exited this development depends on the initial Ethernet work being successful and resources being available.
I hope this answers your questions. I am, of course, happy and anxious to consider any particular requirements you have in mind. I hope you appreciate my reluctance to publish best buys which would only lead to us having to withdraw them and update them to no particular good end.
The group is finding its feet and the shape of the transition is slowly getting sorted out. As usual the meeting was well attended with 15 people from the major centers and DTI.
Work on XXX and FTP has been farmed out to CTIG and FTPIG. These area do not seem to be a problem and there is little argument or issues. It looks like GIFT will spawn an FTAM product and so be an FTP gateway at some time. There is some worry over the state of presentation which could slow things down.
Addressing is one of the hot topics. We attempted to make some progress on understanding the ISO addressing scheme and how this relates to the NRS. An explanatory paper from John Larmouth did not help much. Willy Black, as usual, thought that perhaps we should forget names and stick to a numbering scheme - a view I have a lot of sympathy with. The problem is what to do with the 32 (or is it 40) semi octet address field in CCITT X25 84. There is one standard use which allows the address to be symbolic but this is a bit short on length - the 12 character 'short form' of NRS only just fits in. Long form - well!!! Willy Black again questioned who in their right mind was going to use long form and nobody seemed to disagree.
Mail had an easy ride but is seen as important. A Grey book to MHS gateway will be needed but this is thought to be easy.
JTMP and JTP had a lot of discussion. I was asking about dates and requirements and wondering if we could skip JTMP and go straight to JTP. An idea not well received. Some think 1985 will be the year of JTMP - we shall see.
A useful meeting at which I felt that transition was getting a bit of steam up.
I request permission to run a technical coordination meeting on behalf of the European Academic Research Network (EARN) on February 18/19/20.
EARN is a Europe wide network promoted by IBM and based on IBM protocols. It will have a single gateway to JANET at Rutherford to prevent it undermining JANET. P Bryant is the UK EARN representative as well as its technical coordinator. EARN is committed to move to international protocols. Further details of EARN are in the current edition of Forum. JNT has no objections to the current plans for EARN.
The EARN Board of Directors has asked the Technical Coordinator to organize a meeting of technical representatives from each country. This will constitute 20 to 30 people. The meeting will consider the technical development of EARN and in particular its migration to ISO standards. The meeting will start in the evening of Feb. 18 and finish at mid day Feb. 20. It is intended to use Coseners House for accommodation and meetings. The cost per head will be 62 pounds but this may increase slightly if formal dinner is included.
It is hoped that the EARN Association will finance the meeting completely. The Association is financed by IBM. If this is not possible then each delegate will pay to attend. If finance is not forth coming then it would seem reasonable for Rutherford 'infrastructure' to pay for transport to and from the airport (approximately 160 pounds) as the UK will be spared air fare costs.
The benefits of the meeting will be for the academic community who will use the network.
Division Heads are invited to give permission to use Coseners. They are also asked to allow Computing Division staff to be used to organize the meeting which should be small. Division Heads are asked whether they would still allow the meeting if finance is not forthcoming from the association but that delegates pay for use of Coseners.
Here is the rubbish you need for my lecture which will be a highlight of 1985. I hope it is what you want. I shall need an overhead projector to show my hastily hand written foils which will be scratched out the night before.
The building of a wide area network linking the various universities and the Research Councils' sites started in about 1977. This was a difficult job considering the availability of equipment, lack of protocols and, dare I say it, the ever present lack of cash, to undertake it.
It was optimistically thought that on a site networking would be far easier. This was because there would be no PTT involvement, connections would be faster and more reliable, and there would be far fewer management problems with out the lunatics from 'the other sites' and the costs would be low. How wrong can you be!
In the event the introduction of local area high speed networks has proved to be far more difficult than expected. The reason for the delays are numerous ranging from the lack of resources, through unreliable manufacturers, passing lack of standards and finishing with down right bungling.
None the less there has been progress and this presentation will give an analysis of the past and attempt to predict the future with particular reference to Rutherford Laboratory.
An unremarkable career through a lesser known grammar school in a quiet seaside town led to the requisite number of O and A levels to ensure a brilliant first class honours degree in mathematics at Southampton University. This was gained in the year of 1959 before degrees became two a penny. Dr. Bryant followed through with a PhD in applied mathematics which was a spectacularly unsuccessful piece of theoretical physics. He did learn about computing, rowed and propped up the bar to make it all worth while.
He reluctantly left Southampton to join Harwell vowing never to touch a computer again. This was in the days of head hunting for head hunting sake and he soon became tired of the boredom of nothing and turned to compiler writing for the Atlas computer. He moved to the Atlas Lab when that machine was eventually delivered. Since then he has prided himself on only having had four office moves, one per two reorganizations!
After the end of the happy go lucky days of Atlas he looked after the 1906A operating system and a lot of other aspects of the machine. At this time he first started messing about with networks in the form of RJE stations and the SERC network 'mark one' based on the Experimental Packet Switched Network. After this the Interactive Computing Facility claimed his attention by 'force majeure' and he eventually looked after a gaggle of GEC computers sprinkled like confetti round the universities. These were the first machines to make serious use of SERC net. In fact the network of GECs have had a profound effect on SERC net and encouraged its early development and growth.
Dr Bryant has now stuck his finger in more networking pies. He tries to look after networking on the site and is bruised from one or two local area excursions. He is trying to get a connection to the European Academic Research Network and is the technical director of the network as well as the UK representative.
His ambition- that his next crack at local area networks will be the last and successful.
Dear Tony,
I have just hear that you are out in Kuwait. You LMEN do get to some funny places. How do you do Videotex in Kuwait - don't they have a different alphabet?
Anyway - let me wish you a very happy Christmas and a prosperous New Year except that this will probably not reach you until after the event. It must be nice to live in a warmer climate. New regulations at Rutherford mean that the heating is now turned off at night and the place is ice cold till about going home time! It must be part of the new cold climate for science. You may have heard that we have had our budget cut by Mrs. T - it was student grants or the science budget and the row over grants won.
I think you are now well out of the GEC business. I no longer have anything to do with them. I am afraid Bob Hopgood and Cliff Pavelin finally beat me and everything stopped. Mind you I have come to the reluctant conclusion that GEC will 'never make it' . They tried with their 6300 and UNIX but have made a right balls up and have taken DEC head on and I know who will win that battle. So I am now a good IBM man dealing with IBM comms and IBM PCs. The meals are a lot better! Over the last year I took the decision to spend as much time as possible out of the lab and succeeded well. I spent a lot of time in Brussels on Euro network harmonization work which was a lot of fun. I also got mixed up in ESPRIT which was also fun. My latest travel has been on behalf of the European Academic Research Network which is heavily supported by IBM. So I get a lot of travel round Europe. I have certainly met a lot of interesting people and visited most European countries. I have even started going to French lessons. A lot a people have left the Lab. Phillip and Jonathan went to Beele electronic which is turning out to be a bit of a rocky company and I think they will leave soon. I almost joined then but at my age one cannot take big risks. If they do leave I suspect Beele will collapse quickly. Beele does pay well and Jonathan now spends his money gliding. Goodness knows what Phillip does with his money as his lives very frugally. Shirl is still with us but not all that happy. She has been trying to get back to work with me for a long time and I would certainly like her back. Jacky is also still with us but is no longer on GECs. In fact Shirl is about all the effort left on them. The division was split into two a few months back and GEC support went to the Informatics division and the software stayed in computing. Mike Jane looks after it in Informatics but I feel sorry for him as it is impossible for him to achieve much. Roland is now knee deep in JANET which is heavily underfunded in money and people. I am afraid I have rather lost contact with the LMEN apart from Ken Warner who is now in the central computing at Cambridge. The Engineering department is now knee deep in IBM PCs - and good luck to them. It's a warm feeling to get the strength of IBM round you!
I guess you know Arthur Morse. He told me that someone had recommended me to him and I worked out it must be you. He seems desperate for staff - aren't we all. He was interested in offering me a job or some consultancy. I don't think I am interested in a permanent job but a few weeks consultancy sounds interesting. It is probably the only way I will ever see Kuwait and it seems an exciting place to see. As long as I get the expenses I will be happy. Certainly my finances are in such a state that I am afraid that if I do visit it will have to be expenses in advance! Actually, Arthur has not contacted me for a couple of months so I suspect it has all dropped through. Some you win some you loose. He sent me a lot of bumpf and it all looks interesting. At least you seem to have some finance. Finance is just the problem which is stopping me doing anything interesting. You might have heard of the BRYANT poverty trap - it goes like this. If your project is going well everyone is happy and lets you get on with it. One day you report that the project is a day late so you manager discusses it with you for a day and you are now 2 days late. So next month he says you had better write a report on why you are late and this takes 2 days so you are now 4 days late. And so on until your complete energies are directed towards explaining why you are making no progress. The lab is in just such a condition as are most of my projects.
Anyway I will certainly look you up if I do get a trip out their and thank you for all your advice in the past.
Best wishes
Paul Bryant.
Thank you very much for agreeing to be on the program committee for Euro Networkshop. Perhaps I should fill in a few details for you.
The idea arose from discussions we have had with many people throughout Europe and eventually Peter Linington, James Hutton and myself thought we ought to try and start something. I must say we were a bit reluctant to do this as we thought it might be seen as yet another UK dominated network extravaganza and I feel we have seen too much of that. We are also very keen that the event is run by academics for academics, it would be a pity if Alvey, ESPRIT, the PTTs or government bodies got in on the act as they may well tend to dominate. Hence we are trying to ensure that the event is truly seen as both European and Academic. We also want to foster good relations and cooperating between academics. Although the Zandar harmonization work was very useful I got the feeling that the work became dominated by non academics such as the UK Department of Trade and Industry and in some cases the PTTs. We felt that this was bad and should be avoided.
So what has happened so far. So far it has just been the three of us. We have taken soundings from various places and the idea seems well received. We have put a lot of thought into where it should be. We felt it must be cheap, we want everyone under the same roof to encourage informal contacts, we felt that it should be held on some neutral place to avoid domination. It needed to be central to reduce travel expenses. We did not feel we could stand the basement of the Berlaymont and in any case that would be asking for the ITTT to dominate it. CERN was a possibility but High Energy domination was a worry. Luxembourg was chosen as the best compromise. If we run a successor I guess we can be freer in our choice. Anyway, Luxembourg looks promising.
Our next job was to consider the organization. We felt we needed a program committee with representatives from the major countries so that we could get a program acceptable to everyone and picking up all the important work going on. We are attempting to get some sponsors to put their names on the letter head. This will give us respectability and make it easier for delegates to get permission to attend. We are thus getting together a committee of sponsors which will probably never meet but will have to OK the program. We hope they will supply the after dinner speaker.
So now we have to get the program committee to meet and draw up the agenda.
The attached paper will give you some further information and you will see who the other members of the program committee are. This is just one of our working papers so please do not treat it as the last word on anything. you will see that Peter Linington will be our representative at Coseners House. However, James and myself have appointed ourselves as the secretariat so as not to be left out!
Can I remind you of the details of the Coseners meeting. You should arrive on the evening of Wednesday 16 January. If you are early enough we intend to start the meeting in the pub. The meeting will finish at 16.00 hours on Wednesday 17 January. We will provide accommodation, meals and transport from the airport. I am sorry we cannot provide the air fare. Could you send me details of your flights (via Kom). Also any other comments will be welcome.
I met Hargan Hultzch from GSI last week at the EARN Board of Directors meeting. He tells me he is now helping with setting up DFN. Is this a case of poacher turned game keeper! I am the technical coordinator for EARN and one of my tasks will be planning the migration of EARN to ISO protocols. It's all very exciting and I believe we will succeed. Certainly I hope that Euronetworkshop will help in putting together a coherent European academic network into which EARN will have to be absorbed.
I look forward to seeing you again.
With best wishes
Paul Bryant.
This meeting was held at the request of George Moudiotis of BT as he wanted us to comment on 'block mode working'.
The minutes took all morning to get through and tempers got thin. CTIG was upset by the following:-
CTIG is upset that it is not being provided with information from BT to undertake their job. Technical documents were not provided regardless of the fact that BT was asked repeatedly for them. They were asked to comment on topics without suitable briefing. CTIG is unable to find out what relevant projects were being under taken and were only being informed at the same time as the general public. They even find it difficult to find out where to get document on existing services and wonder how anyone can be expected to use such services. Of particular interest was the technical documentation on the TNA. The original documents obtained a few years ago did not come from 'legal' sources. It is only now that a 'proper' service is being mounted that the meeting was informed of how to get a technical specification and was then expected to pay 5 pounds for a copy.
Overall CTIG feels that it is not being given sensible conditions under which to operate.
[Note. Block mode working is a mode in which data is sent in packets between the PAD and terminal with each packet having a cheek sum. It is expected to be used to ensure reliable communication and will be useful in graphics, banking and file transfer applications].
On more technical matters we considered the block mode working. There were several aspects that gave us cause for concern. Eventually we decided that we did not really understand the spectrum of services provided by PADs which range from ordinary terminals, through block mode working to the JNT asynchronous proposal. We decided to study this spectrum and try and draw some conclusions. We were rather worried over the use of the X3 parameters as the block mode PAD omits some parameters and invents others.
An interesting point is that block mode is attempting to move some of the intelligence from the PAD into the terminal. We thought that there was now a three way split of parameter responsibility. Some parameters were the hosts problem, some the PADs and some the terminals. We felt that this concept needed further study.
Dear ,
At the recent EARN Board of Directors Meeting I was asked to set up a technical meeting for the people looking after and developing EARN.
I have arranged this meeting for February 18 to 20. It will start on the evening of Monday 18th and finish just after lunch on the 20th. This will hopefully make travel easy.
The meeting will be held at the Cosener's House which is a small conference centre in Abingdon run by Rutherford Laboratory. The centre will accommodate between 20 and 30 people which should allow one or two experts from each country. I would like to keep the meeting relatively small as I believe a small meeting is more likely to make progress and allow us to work informally. It may be necessary to limit numbers in which case I propose that countries not yet connected will possibly be limited to one person. I hope to have an RSCS expert from IBM and possible a representative from BITNET.
The cost of the conference centre will 78 pounds per person and unfortunately this will have to be born by the participants since it has not been found appropriate for the cost to be born by the EARN Association. The cost includes all meals, accommodation and taxes. Delegates finding their own accommodation will be charged 20 pounds. I have booked 25 places and the first 25 people to apply and who include payment will get these places. Bookings received over the 25 or not including payment or received after January 28 may have to be accommodated in a nearby hotel which regretfully will be more expensive. Payment should be in the form of a bank draft in Sterling drawn on a UK bank. If you decide not to pay in advance then payment will have to be in cash, sterling cheques, travellers cheques or Eurocheques. Regretfully credit cards are not accepted.
I would like you to nominate your experts who you would like to attend and let me know their travel arrangements so that I can arrange transport from the airport and send them any further literature. May I emphasize that this meeting is designed for technical experts and not EARN Board Of Director members. I attach a form for applications.
Attached is the provisional agenda for the meeting. I am most happy to add any additional items. I would ask that experts in any of the areas come prepared to present their ideas and written contributions would be most welcome.
I hope you will be able to support the meeting and look forward to being able to coordinate the technical progress of our network.
Yours sincerely
Paul Bryant.
IMPORTANT- I intend to introduce each topic with informal short presentations from delegates. I would appreciate people being prepared to talk on topics they feel they can contribute to. These will be more valuable if written contributions can also be provided. I do not intend to keep to the times and these are just for guidance.
February 18.
Arrive. A buffet supper will be available and there will be
an opportunity to get to know each other.
February 19.
9.00 Introductions.
Review of state of EARN - Plans for the rest of the meeting.
Documentation.
How should the existence of documents be publicised? How
should they be circulated and maintained?
Centres of excellence.
It may be appropriate to designate certain people or sites as
'experts' on particular products and provide help to others.
10.30 Services.
NETSERVE is running at Darmstadt, CERN, Spain and Italy and
provides user information services and node management. What
problems are there and what further developments are needed
Directory services.
A directory service for users is needed. What product(s) are
needed, where should they be located and how should they be
maintained? How will this service relate to similar services
on BITNET and other networks?
Mail.
EARN requires a mail service and the MAILER is a possibility.
It is important that it will allow mail to pass through
gateways to other networks. Should we adopt RFS 822 or try
to develop along MHS lines?
14.00 Node management.
EARN needs to decide how routing tables are to be updated,
generated and distributed. How will EARN routing relate to
BITNET ones? Should there be a single master table or should
each country have a separate table?
Network problems.
Problems with RSCS, JES2, JES3, JNET and other products. What
problems exist, what problem or error reporting facilities
are needed?
Accounting and statistics.
Accounting and statistics facilities are needed to satisfy the
PTTs. Can a single program or set of programs be provided to
satisfy every countries requirements.
Management
Network management facilities are needed to ensure maximum
availability. What products exist, how should they be
developed and what new ones are needed?
15.00 Gateways to other networks.
What gateways are to be provided. What possibilities are
there for cooperation? What products need to be developed?
How should inter networking problems across Europe be tackled?
Conferencing.
EARN requires a conferencing system. What system should be
adopted?
February 20.
9.00 User applications.
What applications will be provided? How will these services
be advertised? Should there be an 'EARN Journal' for
advertising services and developments?
10.00 Migration to ISO protocols.
As a result of being allow to use leased lines EARN has
agreed to migrate to the use of ISO protocols. It will be a
job of the technical group to advise on how this should be
achieved. What route should be taken and what time scale
should be adopted?
Future meetings.
Should there be further meetings, if so when and where? How
should the work of the group be developed and monitored
between meetings?
14.00 Depart.
Dear Roland,
I was delighted if not relieved to receive your letter and it is most remiss of me not to have replied earlier. Although I guess I could struggle to you via CERN I have now produced the calling notice for the meeting which I would like you to see and it is too much to type in.
I have attempted some sort of agenda and have put in the topics you suggested and a few others. I don't think it is all that important that it is complete as I intend to run the meeting flexibly and add any new topics that come up. It should serve as a guide and I hope that it will allow the delegates to think about the topics and come prepared with ideas.
I intend to chair the meeting - unless you would like to share the job. It may be an idea if you looked after sessions where I intend to make some technical contributions - like transition to ISO. Incidentally - do you think we need to give a presentation of ISO protocols and X25 or do you think the delegates will be familiar with the topic?
I would certainly appreciate you giving very short presentations on the topics you are working on such as accounting, conferencing and so on. There will, of course, be the usual conference aids for presentations.
Although I have asked for other delegates for topics and presentations my experience is that I will not get much response and only people I approach will contribute. That means that probable it will be you and me driving things - hence my delight at you interest. Who else should I approach?
I would certainly be pleased if not anxious to discuss the meeting further with you. It may be a good idea if you could come over a bit earlier on the Monday when we could go over the final details.
I intend to minute the meeting as I expect it to be the first of a series and I want to make sure that any decisions and actions are carried out. I shall bring two delegates who will also minute some of the sessions. It would ease the burden if I could get one or two other people to help. If there are other delegates coming from Germany perhaps you could twist their arms to help.
May I wish you a happy Christmas and a prosperous New Year? I look forward to hearing from you soon.
Yours sincerely
Paul Bryant.
Dear Roland,
Thank you for your letter.
I hope you will still be coming to Rutherford some time early Monday 18 February. If you still intend coming by train then I had better explain how to get to the site. The train service from Paddington to Didcot is quite good and takes an hour or so. There are three possibilities for getting to Rutherford. First by taxi - there are always plenty outside the station. Second is to catch the Lab transport from the station at 9.15 or 10.15 which is free. At 9.15 look for a door mobile and at 10.15 a bus marked AERE. Make sure the driver drops you at Rutherford and not at AERE Harwell. The third possibility is to let me know when you will be arriving and I will arrange a car.
I have not yet heard from Bethold Pash or Michael Hebgen, are they coming? The current count is 21 people including 2 IBMers from Germany and one from UK. Unfortunately I have been unable to get the RSCS expert from Hursley. I expect a few more yet. Life has been difficult as I had not appreciated just how bad the postal services were in southern Europe and all my careful plans are falling to pieces.
I have amended the agenda slightly and will be sending that out in the next few days. As you suggest, I will give a brief introduction to ISO and X25. The first session should be easy. In the second session I would appreciate your help with a brief presentation of NETSERVE. Could you get Bethold to do it if he is coming or if that's not possible could you do it. Is there anyone who knows anything about the MAILER? I am very concerned that nothing seems to be happening about it and I hope it will be the basis of our JANET gateway. In fact, would it be possible for you to bring across a copy. I tried to get it when I came to GSI but time was against me. Regretfully we still do not have permission to use EARN from our government.
I would also appreciate your help with node management section.
I can cope with the gateway section and I have no doubt that the Swedish delegates will lead the conferencing section although I do not yet see how KOM would fit into the EARN 'way of doing things'. I hope we will find out.
I now have a letter from IBM on ISO and I enclose a copy for you. This, together with the IBM ISO documents should give us a good start for the ISO session which I will cope with.
I am sorry to push so much your way. The problem is that I am very unsure of who will become the experts in EARN. I hope that this meeting will identify them and that at subsequent meetings we can get an expert for each section.
The meeting room will have the usual overhead projector and black boards but I am not intending to lay on a projector unless asked. If the numbers do not rise much I think I can managed a circle of armchairs which makes the whole thing more informal and friendly.
I guess we should have these meetings at 6 month intervals. Thus the next one would be in August. That suggests a north European site. We must attempt to make the meetings pleasant and attractive.
Best wishes
Paul Bryant.