COMMITTEE FOR THE COMPUTING SERVICE
Annual Report for the year 1970-1
Last year’s report noted that 1969-70 had largely been a year spent in negotiations, planning, and preparation for the installation of the 1906A computer. The year 1970-1 saw the completion of the new computer building, the installation of the 1906A computer, the start of the 1906A service, and the planning for the rundown and eventual sale of KDF9. 1970-1 was also the first full year of the Committee for the Computing Service, and this is the first annual report produced by the Committee.
The new computer building and associated offices in 17 and 19 Banbury Road were completed at the end of September 1970, and the Computing Service staff, with the exception of those operating the KDF9, started to move to the new building in September 1970. In 1970-1, 15 Banbury Road was allocated to the Computing Service for office use, and the necessary work of conversion was completed in time for occupation in March 1971.
3. The ICL 1906A Computer
Installation and commissioning of the ICL 1906A started in mid-November 1970. The acceptance test for all equipment except the fixed disc store was satisfactorily completed on 4 March 1971; the fixed disc store passed its acceptance test on 7 April 1971, and subsequently completed an extended test under working conditions on 7 October 1971.
Since the 1906A system was not available for use by the contractual date of 15 February 1971, ICL provided a standby service by installing an ICL 1904A computer on the Oxford site. A service was provided on the 1904A computer from 15 February until the 1906A service was introduced on 8 April 1971. The rather limited service on the 1904A started and ran relatively uneventfully. The introduction of the 1906A service coincided with the introduction of a new mark of the operating system software and this, coupled with the full service being offered, caused many operational difficulties. It was, nevertheless, possible to meet the load of work being submitted to the machine, and to introduce in April a terminal service, using teleprinters to communicate with the 1906A and to introduce in May a ‘cafeteria’ service whereby users could, for suitable jobs, feed their own programs and data into the computer, and take away their results within a period of minutes without any intervention by the service operators.
4. Transfer of work from KDF9 to 1906A
The combination of the heavily overloaded state of KDF9 and preparatory work by the service to ease transfer of programs from KDF9 to 1906A caused users to start transferring work from KDF9 as soon as the standby service started in February. By the end of April the 1906A was running between 250 and 300 jobs per day, and by the end of the period under review all but a small group of users, whose jobs were particularly KDF9 dependent, had transferred their work to 1906A.
A search for a buyer for KDF9 began in May 1971, and eventually agreement was reached to sell it in January 1972 to a company which wished to use part of the equipment to upgrade their own KDF9, and the remainder for use as spares. Unfortunately, the proceeds of the sale (£10,000) had to be returned to the Treasury.
At the end of the previous year the staff of the Computing Service was four senior staff and twelve juniors. During 1970-1 most, although not all, of the additional staff approved and supported, by the Computer Board were recruited. At the end of this period the total staff was thirty-seven comprising fourteen senior staff and twenty-three junior staff.
There were four senior vacancies and one junior vacancy in posts approved by the Computer Board. There were two additional senior vacancies in posts newly approved by the General Board but not the Computer Board.
6. Scale of activities
In January 1971, the last complete month before the 1900 service became available to users, the KDF9 was operated on three shifts by the Computing Service, and over weekends by user departments. The average level of work at that time was 1,411 jobs per week: KDF9 was fully loaded, and the level of work was limited by the service response rather than by user demand.
In July 1971, by which time a substantial amount of work had been transferred to the service on 1906A, the KDF9 was operated on a single shift by the Computing Service and for part of the evening and night shift by user departments: the average KDF9 work load had by then dropped to 795 jobs per week. At the same time the 1906A was operated primarily on single shift with a small amount of evening work; in the last week of the month the computer was switched on for fifty-eight hours. The load on the batch service at that time was about 1,500 jobs per week, and the total load on the 1906A including the terminal service and the cafeteria service was about 2,000 jobs per week.
Thus, in a period of six months, the total load on the service had doubled from about 1,400 jobs per week in January, when the load was limited by computing power available, to about 2,800 jobs per week in July, when there was more than adequate capacity to meet the demand. This gives some indication of the inadequacy of KDF9 for the needs of the University: it shows the level of the suppressed demand from existing users, although the effect of inhibiting new applications of computing to research work will only be seen on a much longer time scale.
7. Future Development
At the end of the academic year 1970-1 negotiations were taking place with the Computing Board for enhancements of the computer, increased backing store and improved graph plotting facilities. These applications were subsequently approved.
Towards the end of the year the University was involved in discussions with the Computer Board and other universities about the setting up of a South Eastern Regional Computing Service to be based at London University. It was not clear what the implications of this would be for the long term future of the Computing Service in Oxford, but the Committee had some reservations on the suitability of the proposed regional service for a university of the size of Oxford.