[CC(88)2] (Ref Nos CHE/24, CHE/24/3)
Annual Report of the Computing Service, 30 June 1986 - 28 June 1987
Software and Service Changes
Computing in the Arts
Performance and Reliability
Appendix A. Staff in post - July 1987
Appendix B. Current User Guides
There can be no doubt at all about the major event of the year, since the period of report coincided almost exactly with the procurement of the systems which will form the University's central computing facilities for some time to come. The Operational Requirement was issued to suppliers a few days before the beginning of this period, and as the year comes to its end we are expecting the major part of the new configuration to be delivered within a matter of days.
The evaluation and shortlisting of initial proposals was complete by September 1986, so that detailed discussions with short-listed suppliers took place during Michaelmas Term. The majority of these suppliers accepted invitations to give presentations explaining their proposals, which were attended by members of the Computing Council's procurement working party, members of the Users' Group, and staff of the Computing Service. There were also a number of visits made by senior Computing Service staff both to the suppliers' organisations and to customer sites.
This last activity included an intensive programme of visits in the United States by three staff accompanied by a representative of CCTA's Project Support Division, in the course of which three major manufacturers and five manufacturers of specialised vector or parallel processing computer systems were visited by at least one member of the team.
Memoranda of Agreement were completed with five suppliers before the end of December, and invitations to tender issued by CCTA during January. Tenders were returned early in February, and the assessment carried out and submitted to the Computer Board for its meeting on 17th March. The University's decision was approved by the Board, and contracts were placed shortly thereafter.
The outcome of the selection process was the choice of a large VAX cluster supplied by DEC as the general purpose computing facility, together with a Convex C1-XP2 vector processor for computationally intensive work.
The VAX cluster will consist of four large 8000 series processors - two VAX 8800 dual processor systems and two VAX 8700 single processor systems - running under VAX/VMS. There will be 36 GB of disc storage, delivered in two phases (15 GB in 1987 and 21 GB in 1988), attached to three HSC70 storage controllers. There will also be six magnetic tape drives, and three printers including an LPS40 laser printer capable of producing 40 pages per minute. This will be, at least for the present, one of the largest DEC configurations in the UK and the largest in the university community.
A wide range of programming languages, libraries and other applications software will be purchased for the VAX cluster from the Computer Board funding, including software for database management, text retrieval, statistical analysis, numerical methods, graphics, and algebraic manipulation.
The Convex Cl was selected to provide a facility for numerically demanding applications. The configuration tendered contains two Cl-XP vector processors, with 1.5 GB of local disc storage and one magnetic tape drive. The Cl has a 64-bit architecture similar to that of the Cray 1, although unlike the Cray the Convex can if desired carry out arithmetic on 32-bit operands at higher speed. Each of the two processors has a power of roughly one-fifth of a Cray 1. Although direct interactive access to the Convex is available, it is expected that for many users the mode of use will be by job submission from the VAX cluster.
One of the strong points of the Convex system is the degree of compatibility with a VAX/VMS environment. The 32-bit and 64-bit floating point representations are essentially the same as the VAX single precision and G-floating representations respectively; the Convex vectorising FORTRAN compiler provides all but a handful of the large range of VMS FORTRAN extensions; and while the Convex operating system is a version of BSD 4.2 Unix, a user shell (known as COVUE, or Convex VAX user environment) offering a substantial subset of VMS commands is supplied.
The Convex was delivered in May, and user trials have begun in mid-June. The DEC equipment will be delivered early in July. Connection of the Convex to the VAX cluster, via Ethernet, is not scheduled to take place until after the cluster has completed its acceptance trials. We can be optimistic that the new services will be available to users before the beginning of next academic year.
One final aspect of the new services deserves comment. This is the provision of graphical and other hard copy output. A small proportion of the total budget has been reserved for the replacement of graphics equipment, and now that the main suppliers are known the selection of appropriate equipment is being made. The present expectation is that a new multi-pen drum plotter will be purchased to replace the ageing Calcomp 1051s, and that there will be sufficient funding for a colour electrostatic plotter to provide plots with colour area fill. Some new colour terminals and possibly a small number of advanced graphics workstations are also being considered.
The decision is also being taken to bring in new graphical software in place of GHOST-80, which has been the principal graphical library in use for many years. The reasons are partly that GHOST-80 can no longer be obtained at low cost (such are the commercial pressures on Government institutions!), and partly that the Computer Board itself is encouraging the adoption of GKS-based software in place of the longer-established software such as GHOST and GINO. It is likely that the service will in future be based on SIMPLEPLOT and UNIRAS, both of which have well-developed interfaces to GKS.
It would be wrong to conclude this introduction without recording thanks to all those involved in the procurement: the University's committees and the University Offices staff who support them, the procurement working party, individual members of staff in the Computing Service, the Users' Group and other users who have contributed their views, and our external advisors from the Computer Board and CCTA. We feel that the new equipment matches closely with the University's aims and aspirations when it embarked on the replacement exercise, and we look forward with confidence to the future.
For obvious reasons this has been a year of consolidation, rather than innovation, in many areas. Computing Service staff have been occupied first in the evaluation of possible new systems, and subsequently in planning the changeover to the VAX cluster and Convex services. Two major growth areas, however, have been communications and microcomputer support.
The infrastructure of the Gandalf PACX 2000 terminal network has been completed, with three nodes located at the Computing Service, the Science Area and the Bodleian Library, and remote shelves in most of the MDX rooms of the voice telephone network. The PACX 2000 nodes and their remote shelves are interconnected by fibre optic links.
Transfer of the majority of existing terminal connections from the older PACX IV switch to the PACX 2000 systems was complete by October 1986. By arrangement with British Telecom, who were simultaneously in the final phase of work on the voice network, the first connections through the new combined voice-data cabling were then established. Work on installing new terminal connections using this cabling began in earnest only in December and January when the telephone system was handed over to the University.
There has been a steady and strong demand for terminal connections ever since. The PACX 2000 network has proved to be a great success, giving a largely trouble-free service with a clean and unobtrusive user interface.
The extent of the Cambridge Ring has been somewhat reduced, in order to arrive at a configuration which is easier to maintain. Several printers formerly driven via the Ring have been transferred to direct VAX connections.
The formal adoption of a workstation policy by the University has contributed to an upsurge in microcomputer support activities. The initial recommendations for those wishing to purchase relatively low-cost IBM-compatible hardware were the Zenith Data Systems Z-148 and Z-158 ranges, for which a direct supply arrangement was negotiated with Zenith. Olivetti and IBM hardware, which had previously been recommended, continued to receive support.
WordPerfect has been selected as the primary recommendation for word processing, and site licences (funded through the Research and Equipment Committee) have been taken out for the IBM PC versions of SAS and GLIM. The pursuit of favourable licensing terms for software seems certain to be a major concern for the entire University community for a long while to come.
At its Trinity Term meeting the Computing Council agreed that a degree of central support should also be provided for the Apple Macintosh family, and Macintosh hardware and software are now available for evaluation in the Service's demonstration centre.
Version 815 of VME was introduced in January 1987. This involved no change in the user interface. The VME system will now be run unaltered until the 2988 is taken out of service in December.
The principal area of software development involved the introduction of the Coloured Book communications protocols. Both FTP (over Yellow Book Transport Service) and JTMP were successfully installed and commissioned during the year. Unfortunately the Glasgow implementation of Grey Book mail, which would almost certainly have been of great interest to users, was not released by ICL and it is now unlikely to be worth mounting during the short period of service remaining.
New versions of FACSIMILE, P-STAT, SIR and TSP were introduced during the year.
Versions 4.4 and 4.5 of VMS were installed on the VAX systems without problem. Major efforts have been made to overcome difficulties with the numerous issues of the JTMP software. There are signs that at last this software is approaching an acceptable standard, and it is planned that a JTMP service should form the means by which jobs are submitted to the National Centres on the new VAX cluster. This position was reached only during the summer of 1987, and no attempt was made to introduce JTMP into full service on the VAX 11/785s.
PAFEC and PIGS were released on the VAXs for the first time, and new versions of FSIS, MINITAB, P-STAT, TSP and the NAG FORTRAN library were mounted. The long-awaited revision of MACSYMA for Version 4 of VMS was also installed. In addition graphical interfaces and device drivers for GIMMS, MINITAB and SAS have been introduced, allowing users to generate graphical output directly from these packages.
Elsewhere in this report mention is made of the decision to move away from GHOST-80 as the main general-purpose graphics library. As a first step in this process, the SIMPLEPLOT library was acquired in June 1987 and brought into service on VAX1 and VAX2, to allow users access to the software before the transfer to the cluster.
With the main systems coming under considerable pressure in their last year of service, the budgeting and allocation scheme has required constant attention. The main area of concern has been keeping the two VAX 11/785s, VAX1 and VAX2, in reasonable balance. A number of groups had to be moved between the VAXs to assist this process. In addition, selected users were accommodated on VAX3 in order to ease the congestion on VAX1 and VAX2.
Three new allocations groups have been established: Agricultural Economics, Earth Sciences, and the John Radcliffe Hospital.
The user registration system has been extended to take in the increasing number of users who draw on Computing Service facilities - terminal access to JANET, the media conversion service, and so on - without making use of the mainframe services.
The regular courses introducing users to the 2988 and VAX systems have continued throughout the year. As was to be expected, once the choice of new equipment was known the demand for 2988-based courses showed an obvious decline. Numbers were as follows:
|Getting Started with EDT||17||212|
|Getting Started on the VAX||20||209|
|Getting Started with ECCE||14||64|
|Getting Started on the 2988||14||70|
Less frequent courses were provided on several more advanced aspects of the mainframe services: Further Use of the VAX, DCL on the VAX, Further Use of ECCE, Further Use of the 2988, SCL on the 2988, Introduction to JANET, Introduction to GHOST-80, and Introduction to Package Graphics. With the exception of the SCL course (given twice), each of these courses was given three times.
A series of lectures on the use of statistical packages, with particular emphasis on Minitab, SAS and SPSSX, was given in Michaelmas Term. This was well attended, and the SAS and SPSSX material was repeated in Hilary Term. A series of seminars in February 1987 introducing database software on microcomputers was extremely popular. Occasional seminars were given on other topics, notably file and mail transfer via JANET and the use of JTMP to National Centres.
The most important new departure, however, was the introduction of regular courses on microcomputer use, in support of the University's workstation policy and the choice of WordPerfect as the recommended word processing package. This development required the installation of Zenith workstations in the Computing Service lecture room, the displaced BBC micro terminals being redeployed elsewhere. Three courses were prepared and introduced between January and April 1987. To date they have been given in the following numbers:
|Getting Started with MS-DOS||9||92|
|Further Use of MS-DOS||7||63|
|Getting Started with WordPerfect||7||81|
The demand for these courses has exceeded expectations, and in the light of the initial few months' experience a much wider range of courses is being planned for next academic year.
With the disappearance of the ICL 2988 service, the opportunity is being taken to revise the classification of user guides for next year. A list of current user guides is given in Appendix B.
Several important developments have taken place, all of interest to Oxford users. The new Cray X-MP/48 at the Rutherford Appleton Laboratory came into service early in 1987, and local users were among the first to be allocated time on the machine for its initial period of service. The second Cray 1-S, with 2 Mwords of memory, was delivered to ULCC in September 1986, and became available to users at the end of October. New Amdahl 5890/300 systems were also installed at ULCC and UMRCC during the year, in each case to provide both scalar computing power and front-ending of the vector supercomputers.
The University was allocated 5% of the ULCC resources and 4% of the UMRCC resources, in what is to be the last year before the advent of the peer review system for National Centre computing. As the tables of usage show, these allocations have been exceeded on most facilities, with the ULCC Crays and the UMRCC CDC7600 proving particularly popular. Despite the latter circumstance, no plans have been made to take up time on any extension of the CDC7600 and Cyber 176 services which may be negotiated by UMRCC; the users concerned are content to transfer their work to one of the Crays, the Cyber 205 or Oxford's new Convex system. The 7600 services at the National Centres, first at ULCC and then at UMRCC, have been a major influence on scientific computing over the years, and it is just a little sad to contemplate their departure.
The laser printer proofing service has continued to be heavily used, despite the introduction of charging for this facility in September 1986. Taking the proofing service and the production of bromide together, the Lasercomp system has no further spare capacity. The hardware is also now six years old, and in view of the long time span of many academic publishing projects it is necessary to begin planning the next generation of equipment.
This topic was discussed at the meeting of remote users in March 1987, and will continue to be a concern until the way forward is clear. One possibility is to plan for an evolution, over a period of time, towards a typesetting service based on the device-independent page description language PostScript It is however clear that this transition would be a fairly lengthy one, since many of the special fonts and characters on which Lasercomp users currently depend are not yet available in PostScript form.
The external advisor's post, which is funded by the Computer Board, fell vacant in November 1986. Since less than a year of the post's tenure remained, it was not felt to be practical to readvertise the position without obtaining an extension of the period of funding. Happily the Board were prepared to undertake a further three years' support of the post, and recruitment is again in progress.
The KDEM team remains stable at four staff, and the service continues to be in demand. More of the scanned texts are now provided to users directly on floppy disc, and the staff now have their own personal computer system to allow them to do this without calling on other Computing Service facilities.
Although new issues of version 1 of OCP have continued, attention is now centred on version 2. The mainframe program was ready for testing by users in August 1986, and during the year it has been used extensively in Oxford and has been offered to a number of UK universities for beta testing. For some applications the new version is some three times faster than the old, as well as using far less workspace to achieve the same results. General issue of version 2 awaits the final edition of the documentation.
In parallel with the rewriting of the mainframe program, a version has been developed for the IBM PC and compatible microcomputer systems. This is to be sold by Oxford University Press. Publication is currently scheduled for the autumn of 1987.
The Archive continues to grow steadily, so much so that consideration must soon be given to the cost in staff time of supporting it, and how this can be provided on a long-term basis. To this end, an application has been made to the British Library for a one-year research assistantship to investigate the use of the Archive's texts by those to whom they have been issued, and also to carry out a pilot study into standards of editing and text encoding which might serve as a model for future development of the service.
In the autumn of 1986 a proposal was put forward for funds under the Computers in Teaching Initiative to develop teaching software for computer-assisted text analysis in humanities subjects. The project would be incorporated into the undergraduate courses of four faculties: English, Literae Humaniores, Modern Languages and Oriental Studies. Project management and the location and support of the programmers working on software development would be the responsibility of the Computing Service.
This proposal was adopted by the University as its bid for the last round of the initiative, and the project was successful in obtaining funds. Two programmers are now being recruited, and work will begin in the Michaelmas Term 1987.
Three lecture courses were given during the year: a course on the Computer in Literary and Linguistic Analysis in Michaelmas Term, one on Text Analysis for graduate students in Modern Languages in Hilary Term, and one on SNOBOL programming in Trinity Term. Seminars on typesetting, publishing scholarly works, and text retrieval using FAMULUS77 have also been given.
Mrs Hockey was invited to give the keynote address at a conference on SNOBOL programming in North Dakota in October 1986, and presented a paper on Micro-OCP at the International Conference on Computing in the Humanities at Columbia, South Carolina in April 1987. Lectures have also been given at a number of universities in the United States and at Birmingham, Cardiff and Newcastle.
Both the ICL 2988 and the VAX systems ran without serious problems, apart from intermittent spells of disc faults on the now rather elderly fixed disc transports. One head-drive assembly was replaced on the ICL 2988, and one complete System Industries Winchester sub-system on the VAXs. Two incidents on the ICL 2988 causing loss of service, one from a power distribution fault in August 1986 and one when the catalogue became full in September 1986, are the only other problems worthy of mention.
A considerable amount of work was necessary on the environmental services during the year, especially during the last few months in preparation for installation of the new VAX cluster. A new power distribution unit capable of supplying both the VAX cluster and the Convex has been installed, leaving the existing ICL power distribution systems in place until the overlap period between old and new equipment has been completed. In addition, the air conditioning system has needed a considerable amount of maintenance work. Regrettably but inevitably these activities have resulted in a number of planned close-downs. It is however expected that the changeover to the new VAX cluster service should now be possible with no more than a day's break in service to any user.
The other services have given rise to few problems. It is noteworthy that with the introduction of the laser printer proofing service on the Lasercomp, typesetting work has expanded to fill the time available on the Lasercomp itself completely. On occasions the backlog of Lasercomp work has been cleared only by Saturday shift working.
The Intermedia floppy disc format conversion service is now also used to capacity, and thought is being given to enhancement of the hardware in order to provide greater throughput.
|Director||A G Robiette|
|Deputy Director||C E Phelps|
|Director's Secretary||D Clarke|
|Group Manager||A R Gay|
|Software Section Manager||J R Douglas|
|Hardware Section Manager||G W Litchfield|
|Group Secretary||M A Turner|
|Programming Staff||M D Austen
J A Burnell-Higgs
C J Hurrell
K A Lewis
M K Malik
D A Miles
D W Rischmiller
J T Thomason
R F Treweek
S E Treweek
|Network Controller||G B Lescott|
|Technical Staff||K A George
A G Marlow
|User Services Group|
|Group Manager||L Hayes|
|User Liaison Manager||C Bateman|
|Computing in the Arts||S M Hockey|
|Group Secretary||R L Turner|
|Programming Staff:||L D Burnard
E M Crutch
C M Griffin
R L Hutchings
J M R Martin
L C Munro
R L D Rees
P S Salotti
E W Taylor
|KDEM Service||G E Cooper
A E Holl
G A Jackson
A B Sabin
|Library Assistant||M E Franks|
|Budgets Office||C Windridge|
|Receptionists||C M Dale
M J Smith
|Technical Typist||E C Hussey|
|Operations Manager||R F Hufton|
|Microsystem Support||S E Evans
A E Lawrence
|Assistant Operations Manager||R I Saxton|
|Shift Supervisors|| D C Hastings
B H Martin
|Operations Staff||J Barker
A P Bourton
G A Fegan
J C Hill
A R Knight
A F Martin
A M Rumble
N J Stevens
D R Saxton
|Operations Assistant||D Williams|
|Administrator||A B Blackney|
|Administrative Assistant||L A Mills|
|Print Unit|| A C Hunter
|General and Cleaning Staff|| A McShane
J M Towner
W J Towner
J H Bunce
A1.1 Introduction to the Computing Service
A4.1 Glossary of 2988 Terms
A4.2 Glossary of VAX Terms
A5.1 2988 Budgeting, Accounting and Scheduling
A5.2 VAX Budgeting, Accounting and Scheduling
B1.1 Oxford Command Specifications
B2.1 Getting Started on the 2988
B2.2 Further Use of the 2988
B3.1 Editing Files (the ICL Editor)
C2.1 Running Algol68 Jobs
C3.1 Running FORTRAN Jobs
E1.1 PDS: Personal Data System
E2.1 BMDP Statistical Software
E2.5 MDS(X) - Multidimensional Scaling Programs
E2.8 CLUSTAN: Cluster Analysis Package
E3.4 GIMMS - General Purpose Geographic Processing System
F1.1 The National Computer Centres: ULCC and UMRCC
F2.1 VAX Network Commands
F2.2 2988 Network Commands
F3.1 Using the PAD
G1.1 Creating and Editing Files using ECCE
G1.2 Getting Started with ECCE
G2.1 VAX/2988 File and Job Transfer
G2.3 The Oxford File Archive
G2.4 Magnetic Tape on the 2988 and VAX
G2.5 EPS 2700 Laser Printer
G2.6 The Qume Printer
G3.3 LASERCHECK at OUCS
G4.1 OCP - Oxford Concordance Program
G4.4 GENSTAT - A General Statistical Program
G4.5 TSP - Time Series Processor
G4.6 GLIM - Generalised Linear Interactive Modelling
G5.1 NAG Library
G5.2 GHOST80 Graphics at OUCS
G5.3 GINO-F Graphics at OUCS
M2.1 CAFS: Content Addressable File Store
M2.5 MACSYMA: An Algebraic Manipulation Package
M3.1 Calcomp-on-GHOST80 on the VAX and 2988
M4.2 Standards for Magnetic Tape Transfer
M4.4 Use of the Monotype Lasercomp Typesetter from Remote Sites
M4.5 Facilities for Reading non-VAX Files on the VAX
M5.2 Trident TT100 Brown Terminals
M5.5 Dacoll M249 Graphics Terminals
M5.6 The Calcomp 1012 Plotter
M5.7 ICL 7561 Orange Terminals
M5.8 Tektronix T4010 Graphics Terminals
M5.10 Datatype X5A Colour Graphics Terminals
M5.11 BBC Micro Terminals
M5.13 Hewlett Packard 7475 Plotter
M5.14 Digital VT220 Terminals
M6.1 Flexible-Disc Conversion Service
M6.2 PDP11 Magnetic Media Conversion Service
M6.3 File Transfer between an Apple and a VAX, using Kermit
M6.4 Cifer File Transfer for CP/M Discs to the VAX, using Kermit
M6.5 File Transfer Using the Cambridge Ring (reading CP/M discs on 380Z)
M6.7 File Transfer between an RML 380Z and a VAX, using Kermit
V1.1 Getting Started on the VAX
V1.2 Getting Started with EDT
V2.1 Using the VAX
V3.1 SAS: Statistical Analysis System
V3.2 MINITAB - An Interactive Statistics Package
V3.3 SIR: Scientific Information Retrieval
V3.4 P-STAT: Princeton Statistical Program
|No. of Jobs||CPU Hours||No. of Jobs||CPU Hours||No. of Jobs||CPU Hours||No. of Jobs||CPU Hours|
The percentage of scheduled productive time (i.e. switched-on time not allocated to maintenance, development or backup) during which the machine is fault-free. Weighting factors are applied to equipment failures that allow the service to continue with impaired performance. Mean Time Between Failures (MTBF) is the mean time between incidents requiring a system reload to take place.
|Ending||Reliability (%)||MTBF (hrs)||Reliability (%)||MTBF (hrs)||Reliability (%)||MTBF (hrs)||Reliability (%)||MTBF (hrs)|
|4-Weeks||Amdahl||Cray-1S (COS1M)||Cray-1S (COS2M)|
|4-Weeks||CDC 7600||Cyber 176||Cyber 205|