ACD Atlas Computing Division Distributed Computing Systems

Jump To Main Content

Jump Over Banner

Home

ACDDCSProjects

Jump Over Left Menu

DCS Projects: University Colleg

P L HIGGINSON

A STUDY OF PROTOCOL REQUIREMENTS FOR MULTI-MEDIA MESSAGES AND INTERCOMMUNICATING OFFICE MACHINES

Jan 83 - June 85

This project aims to study the protocol problems and requirements for multimedia message exchange.

X25 Packet Switching Networks are now well established and there are accepted UK protocols (even if not yet international ones) for Transport Service, file and job transfer. There is an interim JNT mail protocol which is a combination of the well established Arpanet message structure format and the UK File Transfer Protocol. Whilst the Transfer part can handle any data format, the structure is restricted to ASCII text. This is ideal for interim use. but effectively prevents the transmission of other types of data such as graphics, voice, video and facsimile as a part of messages. These defects were known to the definers of the protocol, but were necessary for compatibility with existing procedures and appropriate in an interim protocol.

There is a new set of Teletex Recommendations (ie standards) from CCITT, which are likely to shape future work in this field. In many areas (for example modems) international standards follow from the developments of acceptable prototypes and their use in real situations. Only in data communication do we see seat-of-the-pants design resulting in international standards without any significant implementations being done. For example, X25 and X3, X28 and X29 were produced in this way; they are still being changed at an alarming rate and have serious fundamental weaknesses which there is no hope of correcting.

The Teletex service and protocol, whilst stated to be memory to memory, is oriented towards stand-alone, dedicated machines. The present specification caters only for text transmission although there are plans to adapt it to graphics working. Networks and Teletex customers are planned, and are likely to be an experimental phase during the course of the project.

The CCITT Teletex Recommendation is structured as a number of documents which describe the requirements and general facilities, the basic service, the presentation functions, the end-to-end or session functions and the transport service. As it is one of the first protocols to be designed using the ISO/OSI model; although the model is not yet finalised, there is a clear convergence between it and the Teletex standards. Revisions to Teletex are likely to bring it even closer to the model, and for this reason it is an important protocol to study.

The transport service for Teletex is generally acknowledged to be extremely important and quite adequate, since it is based on the ECMA draft standard (on which much work has been done). The higher levels are inadequate for the long term needs of information transfer, and much work remains to be done. The GILT project is also investigating some of these areas. We have been asked to participate, but so far have lacked the effort to do so.

While early work in facsimile has followed a different direction from Teletex, the future Datafax work should be fully compatible with Teletex. Early Video Gateway work has also been on a different basis, however, the current CCITT thinking is also along ISO/Teletex lines. We have joint activities with BT Martlesham in this area.

This project is intended to show that certain communications and methods are feasible to small and medium sized office systems.

In the initial period the software work has concentrated on completing and enhancing a distributed Teletex implementation which was started under another grant. We have demonstrated ability to send Teletex messages across networks of the Universe type as well as X25 ones and plan further tests to ensure our compatibility with industry standards. We have joined the BT Industry group on Teletex. A Codata 68000 has been obtained to fulfill the small workstation role in the original proposal. and a special high speed interface for connection to ISDN networks is being built. Some work has also been done (in co-operation with the UCL Universe group) on improving the facilities to boot small distributed workstations.

PROFESSOR P T KIRSTEIN

COMMUNICATION PROTOCOLS IN THE CONTEXT OF X25 COMMUNICATIONS PROJECT

Oct 78 - Feb 83

Aims

The aim of the project is to investigate the features and performance of the PITT agreed X25 protocol and of the High Level protocols needed in a computer network consisting of both high speed Local Area Network (LAN) components and gateways to Wide Area Networks (WAN). The correctness, ease of implementation, adequacy of function, independence of level, level structure, dependence on sub-net properties and protocol performance are all being studied.

Experimental Environment

The experimental environment is three Cambridge Rings with a gateway to the X25 PSS and SERCNET X25 networks. For comparison purposes there is also a gateway to the DARPA SATNET and UNIVERSE satellite Networks. A number of minicomputers on a ring can provide protocol translation, performance and availability measurements, and gateway function. About a dozen PDP 11 or LSI 11 computers participate in a distributed way in these functions.

Changes in Direction

Over the last year the specific needs for good facilities for network measurement and control have been recognised. Considerable effort has gone into the provision of such facilities and introducing similar techniques with a follow-on project. Since the start of the project, the standards for document service over such networks have emerged. The protocols of concern have broadened to include Videotex, Teletex, Computer Based Message Systems (CBMS) and Facsimile over such networks. There has been particularly heavy emphasis on CBMS.

Recent Progress

The relevant publications written during the last 12 months are listed in the references; there are numerous additional internal documents.

The project officially terminated in February 1983, so that this report really represents only about 5 months work. Most of our activity was, therefore, devoted to finishing off the work in preparation for new projects. Many of the isolated strands have, therefore, been taken up in new projects.

The work in interconnection architectures was largely completed. We were able to show how to provide the different protocol levels in a completely distributed way, so that the same high level protocols could operate over different low level ones; alternately different high level ones could operate over the same low level protocol engines. Our work with the same file, terminal and mail protocols over the DARPA internet and the X25 networks is discussed in [1] and [15]. This work has been demonstrated by tests with Teletex; here implementations distributed around the UCL Rings. use the same X25 gateways as the other applications.

Naming and addressing have been a problem at several levels of protocol. Our work at the transport and network access level are discussed in [2]. We have become much more active in Mail across different networks. Here again protocol differences have caused us many problems, some of which are discussed in [6]. This work is continuing in a new project under Martin.

The monitoring work has continued, partly with help from this project and partly in the context of project UNIVERSE. One publication in this area is [8]. We have now achieved a good ability to monitor the status of all the network connected to UCL (3 Rings, SERCNET, PSS, SATNET/DARPA CATENET. UNIVERSE), and have a fairly detailed understanding of how such systems should be constructed. Again the work is continuing, but now entirely under UNIVERSE auspices.

As part of the improvement of network reliability, there has been an investigation of automatic bootstrap loading access the Cambridge Ring. While this work has now been largely successful [4], only some has been achieved under the auspices of the present project. It has been continued in a project under Higginson, which started early in 1983.

The theoretical analysis has concentrated largely on UNIVERSE applications [7]. It was clearly that this work would pay highest dividends over the current year. Again it has been continued under UNIVERSE auspices, but will be an independent project from October 1983 under Leung and Kirstein.

Industrial Involvement

The involvement in the Teletex work was with BT and Logica. From their part it was under UNIVERSE auspices; from ours it was partly from this project, and partly under other funding. In any case, the collaboration has been strong. ICL has had a staff member on the group in monitoring activities throughout the year. CASE Ltd has been sufficiently interested in the monitoring work to fund a studentship in this area. GEC is collaborating also with this work - through their involvement is again under UNIVERSE auspices. The CASE studentship from BT has continued the collaboration on distributed Videotex.

References

1. R. Cole et aI, A flexible architecture for protocol studies in a multi-network environment, IFIP83, Paris, pp 401-406. 1983.

2. R. Cole et al, Naming, addressing and routing in a multi network environment, Data Communications 12, 6 pp 175-187, 1983.

3. R. Cole et aI, An internetworking architecture for STELLA and UNIVERSE,', Proc. Conf. Satellite and Computer Communication, North Holland. pp 913918, 1983.

4. J. Crowcroft and R. Hall Distributed Approach to loading in a network environment, International Workshop on Ring Technology, Canterbury. pp 111-122, 1983.

5. P. L. Higginson and R. Cole, Issues in Interconnecting local and wide area networks, Proc. Business Telecom Coni, ONLINE, pp 231-241, 1983.

6. S. Kille ' The interconnection of multiple international mail systems using different addressing strategies, Proc, Symp. Computer Based Mail Systems, Versailles, 1982.

7. S. A. Sorensen, Simulation studies of a UNIVERSE type network, Proc. Conf. Satellite and Computer Communication, North Holland. pp 343-352, 1983.

8. S. R. Wilbur, Aspects of Distributed Systems Management, International Workshop on ring Technology, Canterbury. pp 111-122, 1983.

N MARTIN

MESSAGE SERVICES AND DIRECTORY DEVELOPMENT

July 83 - June 85

Aims

There have been many developments in the area of Message Transport Protocols. The aim of this new project is to investigate some of the more significant ones.

The current system at UCL is based on the JNT Mail Protocol, which uses a general purpose file transfer mechanism (NIFTP) as the mechanism for transporting the messages. It is proposed to evaluate other message formats and message transfer mechanisms. Although the emphasis is intended to be on the latter, the problems of message representation are closely tied in with the transfer of problems and will be considered in that context, as opposed to the difficulties in the area of user presentation. The research is expected to comprise a mix of theoretical and practical evaluations of the various protocols.

  1. The CCITT standard transport protocol Teletex will be investigated as a medium for transporting messages. This work is dependent on the provision of the Teletex protocol at UCL, the Teletex implementation is being completed under an ongoing activity. It is proposed that JNT Mail will be extended to run on Teletex as an initial experiment in this area.
  2. This will later be extended to an investigation of the GILT protocols, which aim to be more sophisticated Message Protocols following the Teletex model.
  3. It is proposed to continue the investigations of the Internet Message protocols based on the new message format which has recently been adopted. In particular the Protocol SMTP (Simple Message Transfer Protocol) which will be implemented and evaluated.
  4. The problems of interconnection by use of telephone networks will be examined. Two protocols will be considered in parallel. The first is the CSNET/Phonenet system developed at the University of Delaware, which fits in very naturally with the mail software currently running at UCL.
  5. The second utilises UUCP and is currently used by Usenet in the USA along with a limited number of European sites engaged in rather ad-hoc communications. It is considered to be a significant pragmatic protocol for Unix sites.
  6. The protocols relating to the BT GOLD/Dialcom system will be examined, particularly in relation to compatibility with other systems. The Telemail protocols may also be considered.
  7. The work of various standardisation bodies in this area is considered to be very important in this context. The work will be described in terms of the models and standards of ISO, CCITT, IFIP, and NBS in particular. This work will be continuous over the period of the project.

An important section of the work will be to determine the problems which result from interconnecting the systems being investigated. The difficulties associated with performing protocol conversions, and defining the services which can and cannot be mapped will be assessed. It is seen as important, to both the end user and each individual system, that it be possible to view the set of interconnected mail systems in a homogeneous manner. This means that each system should in some way view the others as a part of itself. Some of the difficulties of this have already been analysed at UCL. It is hoped to extend this work. In particular work will be undertaken to build sophisticated address and format translation facilities.

A new area which will be considered is that of access control and accounting. The difficulties of protecting and charging for use of an expensive or restricted resource have not yet been analysed carefully in the context of electronic mail. It is clear that these facilities must be available at the mail level. The protocol features necessary to achieve this will be evaluated. Experimental implementations are planned.

It cannot be over-emphasised that these problems cannot be completely resolved by purely paper studies: only by implementation exercises, backed up by practical evaluation, can the complete problem be studied.