In June 1973, the Engineering Board set up a Working Group to clarify future engineering computing requirements. The report of this Working Group was considered by the Board in December 1974 and its recommendations were strongly endorsed. The Council subsequently (February 1975) supported in principle the recommendations and invited the Engineering Board to arrange the preparation of detailed proposals for implementation of these recommendations.
As a result, the Rosenbrock Technical Group was set up by the Engineering Board in March 1975 to consider the computing needs of engineers whose research would be supported by SRC in the next five years. The Technical Group recommended a coordinated provision of an interactive facility based on central facilities and local multi-user systems with management of the facility from Chilton (Atlas Division of Rutherford Laboratory). It felt that current industrial work used design methods which did not fully exploit the opportunities of man-machine interaction. It saw a need for university research of very high quality in the next ten years, devoted to finding new design procedures specially adapted to CAD. It was rea1ised that a hardware provision alone would not provide a useful facility. Consequently, it was recommended that Special Interest Groups be formed with manpower provided to establish a uniform software base in each area.
The recommendations of the Rosenbrock Report were accepted by the Board and, as a result, the Interactive Computing Facility of the Engineering Board was set up. Central facilities were provided by DEC10 systems at Edinburgh and UMIST, 23 PRIME and GEC systems were installed in university departments and the facility was managed centrally by the Rutherford Laboratory. It was recommended that 29 direct man years of effort should be put into the facility in the first year rising to 57 man years by the third year.
A significant proportion of the DEC10 at Edinburgh and smaller amounts of the other resources were used by the Science Board and it had a member on the Interactive Computing Facility Committee who acted as its representative.
At the time of the Rosenbrock Report, it was only just looking possible for good interactive facilities to be made available via multi-user mini systems. Consequently, the decision to go for a large facility based primarily on such systems was adventurous. In retrospect, it has proved to be a sensible decision and greatly improved the interactive facilities available to both Engineering and Science Board users.
By putting a significant amount of manpower into the project, progress was fast and the facility was soon giving benefits to the community.
The Rosenbrock Report considered the possibility of providing the interactive service based on local single user systems. It did not believe that such systems were cost effective at that time nor did it believe that it had the ability to provide a similar service based on 60 or more single user systems. The lack of network access to such systems would have made it impossible to provide central management. Access to a number of multi-user systems would also be a problem, and it was largely due to the manpower provided by the Engineering Board and the requirement to support the multi-user systems that SRCnet evolved. At that time, SRC only had star networks around the central batch facilities at Rutherford and Daresbury.
In the longer term (past 1978) the Report anticipated that improvements in single user systems would make them a viable alternative to the multi user systems as long as a national computer network was available to allow user intercommunication. Without this, it saw that the advantages of an integrated system would be lost.
In the period 1977 to 1979, the Interactive Computing Facility did explore the possibilities of single user systems on the market. Systems were purchased from IBM and Tektronix for evaluation. The general conclusions were that they were over-priced, under-powered and lacked the necessary software to make them a viable alternative to the multi-user systems at that time. Even so, a number of packages were developed in the architecture area on the Tektronix 4051 systems and interactive use of APL on the IBM 5100 systems did find acceptance in a small community.
In summary, by May 1979, the Engineering Board had a very effective interactive facility based on multi user minis and central facilities installed. Most of the systems were interconnected by either DECNET, PRIMENET or the embryo SRCnet. Within the next 18 months, it was anticipated that all the systems would be networked together via SRCnet. Six Special Interest Groups had been established and a significant body of software installed.
In June 1976, the Engineering Board Computing Science Committee recognised the importance of the Distributed Computing Systems research area by appointing a panel to consider what action, if any, was necessary to encourage, coordinate or direct research into distributed computing. In October 1976, the Panel recommended that a coordinated research programme should be established and that additional funds should be sought for the programme. As a result, the community was consulted and a proposal to set up a coordinated programme of research was made and accepted by the Engineering Board in June 1977. Coordination for the programme started in January 1978. A DCS Panel reporting to the Computing Science Committee was set up to manage the programme. Bob Hopgood was appointed academic coordinator and Rob Witty acted as Technical Secretary.
The primary objective of the programme was to seek an understanding of the principles of Distributed Computing Systems (DCS) and to establish the engineering techniques necessary to implement such systems effectively. A more general objective was to achieve results of practical value to UK industry by directing research to a key area for the future. Within the context of the programme, a DCS was considered to be one in which a number of autonomous but interacting computers cooperated on a common problem. Motivations for the programme included the importance of distributing processor power to where it is required in order to minimise communication bandwidths and the belief that the cost of microprocessors would decrease and their performance increase so that tasks could be more economically performed on a distributed set of processors.
Among the early grants awarded was one to Professor Coulouris at QMC to investigate the distributed system requirements for effective man-machine interaction. The project was motivated by the concepts of the paperless office and the integrated personal information processing system. The main aim of the research was to identify the hardware and software requirements for the development and operation of highly interactive information processing systems based on distributed single user systems with the automated office as the test bed for the research. Visits to Xerox PARC had indicated the viability of the research. At that time Xerox had established a similar distributed system based on their in-house Alto single user workstations. The QMC Project was aimed in part at building the necessary hardware as none was available on the market. Experiments with LSI-11s early on showed that they had insufficient power and a later design involved PDP11/34s.
Other projects related to the concept of a distributed system of loosely linked single user systems were a distributed filestore (Keele), local area network servers (Kent) and software engineering workstations (Professor Hoare at Oxford).
There was a serious need for a high powered single user system to act as the major vehicle for the coordinated programme. Early on, LSI-11 based systems were purchased but it soon became clear that they did not have the power to do the demanding tasks required nor did the DCS Panel believe it was sensible to initiate its own hardware production in this area as it was likely that such systems would appear on the market.
In summary, by May 1979 the Computing Science Committee (chaired by Professor Rogers) had a thriving Distributed Computing Systems coordinated programme which needed a high powered single user system in order to achieve results from some of the grants already awarded.
In 1978, Rutherford Laboratory had two IBM 360/195 systems providing a mainframe service for all four Boards of SERC. These systems were obsolete with a locally produced ELECTRIC system for interactive editing and job submission. There was an urgent need to enhance these facilities in such a way that more resource was made available for the ever increasing number of logged-in users and to provide a path forward in replacing the 360/l95s.
There was pressure at that time by ICL to provide the Laboratory with a 2900 system as an alternative front end to the 360/l95s. As there was little sign that ICL would be producing replacements for the 360/l95s, this was an unattractive way to go forward and the decision was made by the Laboratory to purchase an IBM 3032 which started the process of replacing the 360/l95s eventually by an IBM 3081 and a Fujitsu M380.
The Laboratory was interested in extending the Office Automation facilities on site at that time. An ICL 2904 was used for processing SRC Grants and Awards and it was feasible that ICL might have a suitable offering in the Office Automation area. As a result, a dialogue with ICL took place during 1979 which discussed possible ICL offerings particularly those of relevance in the scientific office systems area.
By May 1979, the ICF and the DCS programmes had a need for a high powered single user system. Now that the networking was in place it was feasible to consider such systems as a viable way of providing local interactive facilities. In the case of DCS, some large grants were held up due to the lack of such systems.
The final point was that Rutherford were having an ongoing dialogue with ICL concerning possible future products. Such a dialogue was also taking place with GEC concerning a 32-bit version of their multi-user systems.