Produced by the Computer Board And Research Councils' Joint Network Team and Network Executive.
It is with regret that I have to announce the imminent departure of Barrie Charles from the JNT and Keith Mainwaring from the Network Executive. Barrie is leaving after 5 years as one of the mainstays of the JNT to join Digital Equipment as UK Educational Marketing Manager. Keith, who has played an important role in the establishment of the Network Executive, is joining Logica. We wish them both well.
Interviews for the previously vacant post in the Network Executive will be held this month. It is hoped to advertise the newly vacated posts in due course.
Peter Linington
The Cambridge Ring course held at the University of Kent in April was highly successful and substantially over-subscribed. Another course has therefore been arranged for 24-26 September at a cost of £150 for academic and £225 for commercial participants. Further details and registration forms are available from:-
Mrs Shirley Wood, Computing Laboratory, University of Kent at Canterbury, Canterbury, Kent CT2 7NF
Barrie Charles
An agreement was concluded in late March which will allow DEC to supply and support the Coloured Book software developed by UWIST and the University of St Andrews for VAX/VMS. Consequently, the JNT is running down its maintenance contract with UWIST and from now on those requiring the standard software on VAX machines should direct their enquiries to DEC. The development and support costs will now fall on the users of the software although specially reduced prices have been negotiated, with Digital for the academic community. All existing users should by now have been informed of the situation and registered their systems with DEC.
For more information, please direct enquiries either to your local DEC account manager or to the Digital Classified Software Hot Line on Reading (0734) 853071.
Barrie Charles
At the time of writing, we have still not received a field trial version of GEC's type III switch software, which is the product we intend to run on JANET's switches, although delivery is now imminent. As a result, and in order to reduce the possible delay arising from this late delivery, we have begun investigating the management facilities in the software using the type 11 product which is currently mounted on the switches at UMRCC and Belfast, as well as in a large number of CPSEs. In particular, work is in hand to develop an operator's console support package using a microcomputer terminal emulator. In the meantime, we will continue to run the old SERCNET switch software using the package developed by Computing Division at RAL.
We have debated at some length the merits and demerits of various addressing schemes for JANET, but in the end the Executive has had to make a somewhat arbitrary decision as to what scheme to use.
JANET will use X25 addresses of 12 digits, (with 2 further, optional, sub-address digits). Each campus or site will be offered up to 100,000 addresses for its own internal use. This means that if sub-address is utilised, 10,000,000 separate entities can be addressed. The structure of the address will be:-
0000rssXXXXX where r = region digit ss = site digits within a region XXXXX = for use within the site
At a time, as yet unspecified, but likely to coincide with the final installation of GEC's packet switch software on all switches (at present scheduled for 1Q85) all 12 digits of the address should be supplied in the called and calling address fields in packets associated with setting up network calls.
A number of sites, particularly those who have been connected to the UMRCC PSE, have already adopted this scheme. Sites should contact Dr I L Smith of the Network Executive for further information.
We have been asked to clarify the situation on gateways between JANET and X25 networks on campuses.
Basically, the Executive has decided that these are not mandatory - ie that we do not require a site to have a gateway to control access to JANET.
A number of sites have expressed concern as to what effect this may have on their ability to control and account for use of PSS via JANET. This is dealt with below.
Our aim in setting up JANET is to provide connectivity between users on University campuses, at Research Council sites and at a limited number of other specific locations associated with the first two. Thus, once fully operational, JANET should provide a means of access which will satisfy the requirements of a large number of people in the academic community.
However, there is still a requirement, and a growing one at that, to access other sites within the UK and particularly abroad, and the Executive has been charged with providing gateway facilities on to BT's PSS service in order to enable this access to take place.
There have been a number of discussions and proposals put forward as to how this access could best be provided - in particular, there was useful discussion at the IUCC Management Conference at Cambridge in March and at Networkshop 12 at Bath in April on the subject.
The JNT and Executive were proposing a scheme in which accounts on the JANET gateway(s) were identified by calling Transport Service Address, with calls for which charges would be incurred only being accepted for "trusted" addresses. Sites would then be billed for usage against those "trusted" addresses. This generated considerable anxiety among many delegates at both meetings as to how to make appropriate charge back to the actual users, and what facilities would be required to support this activity. There was strong pressure from a number of delegates Lo provide much more sophisticated facilities centrally on the JANET gateway, with the Executive countering that they did not have the manpower to administer huge numbers of accounts.
A compromise proposal was made at Networkshop 13 that the Executive should provide detailed call information (either on hard copy or as a file available for collection using FTP) to enable sites to sub-divide accounts. To make full use of this facility, a Network User Identifier (NUI) would need to be added to the TS calling address. The Executive would then be responsible for setting up the basic accounts, and the call information would allow the generation of usage statistics on a per user basis. Detailed proposals along these lines will be prepared in the near future.
A CPSE gateway would give very much the same level of control as that outlined above, and would generate account records which would still require post-processing and do not currently contain NUI information. Consequently, there seems little to be gained by retaining a CPSE gateway simply to control site access through JANET to PSS. The retention of such a gateway is likely to prove a positive hindrance to easy working over JANET (particularly for X29 users) and may lead to further problems in any future transition to ISO standards.
Ian Smith
The need for a coherent plan for including microcomputer based equipment in the academic community's Open System networking strategy has been apparent for some time. The enormous efforts that were devoted to protocol development on mainframe and minicomputer systems meant that very little time was left for examining the problems of bringing Open System working to this very important area. Recently, however, several factors have led to increased activity.
The community's development projects for the larger machines, whilst not complete, have now produced a wide range of products. The Computer Board has stimulated interest in microcomputer networking with its initiative on the use of computers in teaching which favours the use of individual workstations. Finally, the workstations themselves have grown more powerful and it has become feasible to consider mounting more complex networking packages.
The main means of connecting microcomputer workstations to networks so far has been by providing software to make the micro emulate a simple terminal. A normal asynchronous line permits terminal access to remote systems and by switching input and output to file, a crude file transfer capability can be provided. One of the main requirements for improved network access is to provide a more reliable, easier to use, more manageable and possibly faster file transfer mechanism.
The London Network Team (on behalf of the JNT) organised a meeting to discuss this problem in June 1983 and a second meeting was held in May 1984. The feeling of the meetings was that the direct connection of workstations to networks such as Ethernet or Cambridge Ring was not yet cost effective for the low end of the microcomputer range and that in many cases, high bandwidths were not essential. Two main alternatives were foreseen.
The first used the Asynchronous Transport Service (ATS) to provide communications over an asynchronous line to a terminal concentrator which would switch calls at the Yellow Book Transport Service level onto the network. High level protocol communication using Coloured Book software could then take place directly between the micro and the remote host.
The second alternative was to use the micro manufacturer's proprietary network (ECONET, CHAINNET, QLNET etc) and to gateway this onto the Open network. This alternative is most suited to micros in clusters and offers potentially very low connection costs. The cooperation of the supplier is often essential in this method because of the dependence on proprietary protocols. For this and other reasons, it is to be left to the suppliers to take an initiative in this area, possibly in collaboration with specific universities. The JNT has written to major UK suppliers informing them of these networking alternatives. Acorn, for example, is already actively pursuing some of these ideas.
The production of ATS and Coloured Book (particularly File Transfer Protocol) implementations for micros is seen as requiring a major community initiative, hopefully, again, with the active collaboration of suppliers.
The main reason for the delay between the meetings referred to above, was for the production of proposals from Camtec Electronics for implementations of ATS on the JNT PAD and their other terminal concentrators. The JNT has now placed a contract with Camtec for this development which should be completed in 1Q85. ATS PADs will be fitted with a new memory card to provide extra space to hold code and buffers. It will be possible to upgrade existing PADs as well as to buy new ATS PADs.
Westfield College are already producing an ATS/FTP product for CP/M systems and the JNT is discussing possible developments with other sites. The aim will be to produce a range of products for the more popular micro systems. An Asynchronous Transport Networking Implementor's Group (ATNIG) has been set up to keep a register of implementations and discuss problems. The chairman is Alan Chambers (0272-423478) and the secretary is John Seymour of the London Network Team (01-405-8400). The first meeting was held on 26 June.
In many respects this initiative is overdue and the strategy outlined above is interim pending the direct connection of micros to Open System LANs at competitive prices. Nevertheless, it is essential that something is done now and the products that will become available in 1985 should provide very effective facilities to tide users over this difficult period.
Anyone interested in further information or in getting involved in implementation should contact John Seymour or the JNT.
Barrie Charles
The MACE is at last beginning to reach the final stages of development and testing. The software for the MACE has been tested and accepted with only minor problems which are now rectified. The performance of the software looks to be good and we shall soon have a high speed front end interface on which host developments for VAX and GEC can be completed. The MACE will support 52 simultaneous channels and achieves data throughput rates of 800 Kbps with 4 Kbyte packets and 500 Kbps with 1 Kbyte packets. Acorn are not quite ready to run acceptance on the final hardware but these test are scheduled for early July.
The host interface adapters for Unibus and GEC DMA have also passed their acceptance and I have one of each model in my possession.
So that we waste as little time as possible in moving to production after final acceptance of the hardware, the JNT held a meeting with Acorn representatives during June and came to an agreement on production and support for the initial batch of MACEs. The agreement covers support of both software and hardware and both sides are keen to see the MACE become a successful component in our range of CR82 products. The JNT has now placed an order for 20 MACE/host interface units on behalf of university/Research Council customers and to provide a small central stock of spares. The order is conditional on completion of the hardware acceptance tests. The current production price for a MACE and host interface (GEC or UNIBUS) is £3200 (excluding VAT). The Computer Board will meet the support costs for the units which it has purchased for the first year.
I know many users have been waiting a long time and for this I apologise but it is nice to know that at last deliveries will soon start. I must also apologise for the doubts expressed over Acorn's capabilities in the last issue of Network News which are at last being proved unfounded.
If any of you are not covered by the present JNT provision and want to order MACEs, then I suggest you contact Jim Craigie of the JNT.
Ted Owen
Network News is published every four months and provides news for those interested in networking activities in the Research Council and university community. It is also circulated to polytechnics and interested manufacturers. Requests for copies or items for inclusion should be addressed to the editor.
Editor: B J Charles, Joint Network Team
