With this first issue of Forum in the SERC financial year 1987/88 it is timely to reflect on the results of two years of direct charging for the use of the mainframe computing facilities located at the Atlas Centre, RAL. For those unaware, direct charging was introduced in 1985/86 principally to put the consumers of the computing resources (i.e. mainly the Boards of SERC) in more direct control of the level of resources to be made available. The model chosen for the costing of this directly charged service was designed to cover the annual operating costs plus a provision of £1M for the upgrading of equipment. In practice the funding of the service has been by the provision of individual allocations by the Boards and Divisions of SERC. The funding has proved adequate to cover operating costs but not nearly sufficient to provide for the anticipated upgrade requirements. The Central Computing Division is sensitive to the requirements of its customers and is closely monitoring the level of service available as demand grows. All options to maintain performance under the current funding limitations are being actively investigated.
The year 1987/88 has seen the introduction of the national supercomputing scheme for access to supercomputing (vector processing) facilities at the University of London Computing Centre, the University of Manchester Regional Computing Centre, and the Joint Research Councils' Supercomputer Unit at the Atlas Centre. Access to these facilities is via peer review, and it will be necessary for any project seeking access to demonstrate the need for vector processing techniques. The user community for the Atlas Centre Cray X-MP/48 service is growing at a rate dependent on the the due processes of peer review. This facility is funded by a central allocation but is partially dependent on the mainframe facilities whose funding is subject to the vagaries of the allocations provided by the Boards.
It is interesting to compare the levels of activity for the main services provided over the last two years. In terms of the accounting units the main batch service (MVS including MVT in 1985/6) has shown no change. The CMS service (mainly interactive) has shown an increase of 16% in 1986/87 compared to 1985/86. This is attributable to real growth in the number of interactive users and the movement from MVS to the use of SLAC Batch running under VM/CMS. The impact of the Supercomputer service on this situation will no doubt be visible in the figures for 1987/88. Within CMS we include UTS figures. Use of UTS has declined by some 5% in 1986/87. Are there any frustrated UTS users out there?....
The Divisional Structure of Informatics was reorganised in November 1986 partly to accommodate some changes in senior staff and their responsibilities and partly to align the Division to the structure envisaged for the IT86 follow-up programme to Alvey.
Rob Witty had until July been running the Software Engineering Group in the Division and had acted as Deputy Director to the Alvey Software Engineering Programme. From July until recently, Rob has been the Alvey Software Engineering Director. In April, he leaves for a well-earned sabbatical at Xerox PARC in California.
To enhance the support for the Alvey Infrastructure Programme working for Millbank, the Laboratory has agreed to let Ken Hartley spend half of his time there working directly for Laurence Clarke, the Deputy Director of the Alvey Programme who is also responsible for the Alvey Human Computer Interaction (HCI) Programme.
As a result of these two changes, the Division has changed its Group Structure to:
I am pleased to say that Mike Jane has temporary promotion to Grade 6 in recognition of the additional responsibility.
The main changes from the previous organisation are that Cliff Pavelin has added Software Engineering to his IKBS responsibilities but has passed responsibility for HCI activities to Ken Hartley. As a result, Ken is responsible for all the activities which look like becoming the main infrastructure in IT86. This regrouping is similar to what is proposed for IT86 with IKBS and SE being more closely associated and HCI being an area which needs to be tackled across the whole of IT86.
The new Group run by Mike Jane has responsibility for the earlier Interactive Computing Facility and Single User System programmes which have been amalgamated into a single Engineering Computing Facility. This Group is also responsible for the new Engineering Board Transputer Initiative which is just getting started. To manage the activity, an Engineering Computing Facility Executive has been established which is the main link between the Laboratory and the Computing Facilities Committee and its Advisory Groups. Both Technology and Informatics Divisions are represented on the Executive with Geoff Lambert responsible for the organisation.
Staff below the management level have tended to retain their existing posts although Peter Hemmings has moved to work in the Alvey Infrastructure area taking over some of Cyril Balderson's duties. Cyril, meanwhile, is spending time helping Ken Hartley in his Millbank duties. Peter Kent is now responsible for answering all queries concerning Single User System purchases. Mike Claringbold has taken over responsibility for the operations and user support activities of the facilities run by the Division.
At the outset of the Alvey Programme, the IKBS (Intelligent Knowledge Based Systems and SE (Software Engineering) Directors set up computing facilities for the academic part of their combined areas, and these were subsequently joined by MMI (Man Machine Interface, more usually called HCI: Human Computer Interaction these days). These facilities are managed as an Infrastructure Project by RAL on behalf of Alvey.
The Alvey Computing Infrastructure Project provides standard hardware and software systems for use by grant-supported researchers from the above areas. The hardware is in the form of Multi-User Minicomputers (MUMs) and Single User Systems (SUS). The software is a selection of programming languages relevant to the research activities and running under a number of flavours of the UNIX Operating System.
All the MUMs are connected to the wide-area academic packet-switched network JANET for ease of access by collaborators and for central systems support. SUS can also be connected to JANET in some circumstances, but they are more usually connected to Ethernet Local Area Networks (LAN). A LAN connection is necessary where the SUS is to be connected to a central server.
A mainframe UNIX service is available on the RAL ATLAS 10 computer using the Amdahl UTS (Universal Timesharing System) system, which is also accessible through JANET.
With the VM/370 Operating System, the Atlas 10 is able to provide a Virtual Machine, with 16Mb memory, running the Amdahl UTS, which now provides an implementation of the AT&T UNIX System V. Those who have used this system before will recognise this is a recent update from the original UNIX Version 7. A major facility available to UTS is an IBM 4250 type-setter printer which, in conjunction with a version of the type-setting language titrof produces publication quality masters. The IBM 4250 provides a resolution of 600 X 600 pixels per inch.
The aim of the Alvey Infrastructure Computing Project is to provide and support a range of software of special interest to the academic Information Technology community.
The general programming languages C, Pascal and Fortran 77 are available for all the machines. Also available are a number of implementations of the IKBS languages Prolog and Lisp in addition to the mixed-language environment Poplog. For more information on these language implementations and availability, call Mr Tony Lucas of RAL.
In the Software Engineering area, the standard ML programming language has been implemented at RAL, and details of this and other SE software are available from Dr David Duce.
The MUMs were installed in Departments or Computer Centres at some Universities where there is Alvey supported activity in Information Technology. Each of these "Sites" provides a Site Manager funded by the Project who looks after operations and systems administration on the machines; user support (accreditation, general advice, fault reporting etc) for users; machine performance reporting and Site Contract liaison with RAL.
SUN workstations are purchased centrally by RAL on behalf of SERC and Alvey academic grant holders. Holders of Alvey-funded Sun Workstations are encouraged to contact local Alvey Infrastructure site managers for assistance in their systems administration or for archive facilities on the MUMs. See the table at the end of this article for contacts.
The Atlas-10 Mainframe is run by the RAL Central Computing Division with Systems Support coming from colleagues in the Informatics Division.
The MUMs funded by Alvey are provided as a general facility where there is a community of Alvey research workers. There are three categories of authorisation under which the facilities can be used.
Holders of SUS, since they are provided against grants, clearly need no additional authorisation for their own machine.
Application for use of Mainframe resources (UTS) should be made in the first instance to Mr Peter Hemmings at RAL, for discussion of requirements, access and authorisation.
