While the year was a period of stability regarding the hardware of the Central Computing Service, it saw major changes in other areas. In April, direct charging was introduced, as recommended by the Macfarlane Working Party, and the IBM MVT operating system was replaced by the more modern MVS. Despite this major upheaval, the Service continued to run large amounts of work for users (Table 5.1).
| Funding Source | Batch | CMS |
|---|---|---|
| ASR Board | 16685 | 1140 |
| Engineering Board | 76818 | 2729 |
| Nuclear Physics Board | 657793 | 9824 |
| Science Board | 103206 | 2608 |
| External | 21581 | 645 |
| Total | 876083 | 16946 |
Note: Batch AUs are calculated as MVS AUs + MVT use converted at 1 hour = 100 AUs.
Direct charging is intended to make the Central Computing Service more responsive to the needs of the Council's research programmes. Each Board is required to pay directly for its use of computing resources and is offered a generous discount for making firm financial commitments at the beginning of each year. Rates are set to recover operating costs and to fund a capital replacement programme but initial indications are that revenue. while covering running costs, will be insufficient to cover the replacement programme.
The long-heralded change to MVS was the culmination of two years preparation. MVT had not been supported by IBM for many years and was incapable of taking advantage of modern hardware. The RAL MVT system had been heavily modified over its 14-year life to cater for the needs of the specialised user community and corresponding changes had to be made to MVS and its spooling sub-system JES3. Basic system modifications and a complete re-write of the accounting and batch graphics systems were required.
Release 3 of the Control Program (CP) part of the VM operating system was installed in January. This allowed the introduction of the IBM database system SQL/DS to enable CMS users to query information relating to the tape and disk management system TOMS. SQL/DS is also available for CMS users to set up and query their own databases.
As part of a collaboration with CERN and other high energy physics centres, much work has been done on unifying the RAL and CERN VM/CMS systems. This has included an exchange of CP and CMS modifications, the production of a set of common fund EXECs and macros which can run at each site and the installation of RAL facilities at CERN and vice versa. Tests have been carried out on an improved batch system for CMS which originated at SLAC. SLAC Batch has been interfaced to the CMS accounting system and will be a replacement for the existing Batch Monitor on which it is based. Further work is needed to interface it to TOMS before it can be utilised fully as a batch system with access to tapes.
An important change was made to Atlas 10 software. The VM operating system runs under the control of a hypervisor whose main function is to simulate in microcode the action of several privileged machine instructions. It was discovered that the Atlas 10 was spending too much time in the hypervisor and it was realised that the hypervisor functions could be carried out much more efficiently if they were performed by CP. The hypervisor was replaced by the modified CP resulting in 20-30% more CPU power available to user jobs in the MVS virtual machine.
Many other enhancements and additions to CMS commands and facilities were made, including installation of the IBM Document Composition Facility (DCF/SCRIPT). Initially, this was used by the office automation system, PROFS, but was later made available to all users as a text processing system with interfaces to the Xerox 8700 laser and IBM 4250 electro-erosion printers.
Because most effort during the two-year conversion programme had been directed at implementing local requirements in MVS, such as the tape management system TOMS, the accounting/resource control system and the dataset protection system ACF2, the initial production version was several releases out of date. Subsequent effort has been directed to bringing the system up to the latest IBM release. Further development was required to run a local MVS system on the IBM 3081 so that this machine could be utilised outside prime shift when there is very little CMS activity. This went into production at the beginning of December.
Another development brought the Masstor M860 storage device into service under the control of the data management system ASM2. It was used for taking back-up copies of the MVS disks and, from 28 October, for the migration of datasets which had not been used for a specified period. Datasets in the M860 are retrieved automatically within a few seconds when referenced by a job without the user having to take any specific action. This migration applies only to complete datasets. To allow archiving of individual members of user libraries, the program PDSMAN was installed. It provides user-controlled archiving to and retrieval from magnetic tape, a clean-up and space re-use system, and facilities for listing library and archive members.
The performance monitoring and analysis systems were modified to reflect the move to MVS and the subsequent introduction of a local MVS system on the 3081. A new version of the statistical analysis package SAS was installed. This is used to analyse results from the CMS and MVS monitoring tools and to produce reports and graphs which are used to spot performance problems and help with decisions on corrective action.
The Program Advisory Office handles queries from users via electronic mail, telephone and visits from users. Under normal circumstances, the Office handles about 35 queries a day. The move to MVS saw this rise to 80 per day, many of which were of a novel character. This required a new system for handling user queries which allows a team of experts to work independently on separate queries. In addition to dealing with specific problems, there has been a need for user training on MVS. User Support staff have written a primer and run several conversion courses. User turnover is high and there is a continuous training requirement. This year saw the introduction of a new course covering CMS and MVS. The course gave attendees a chance to practise their newly gained knowledge with supervised tutorial sessions accounting for a considerable percentage of the course. The course was acclaimed as a success and has been run four times for a total of over 60 students. The material developed for the course has been made available to existing users.
Once again, there has been a large increase in the use of office automation based on the IBM PROFS (PRofessional OFfice System) product installed at RAL. In the 1984 Annual Report, the number of users was noted as having grown from 40 to 250; the number has now passed 400. There have been significant developments in the provision of access to and from PROFS over various networks accomplished by the use of PROFSNET. There has also been development work on the use of IBM PCs for office automation and considerable effort has been expended on training. A new user-friendly front-end system has been provided between PROFS and the information retrieval system containing SERC Financial Memoranda and Conditions of Employment, Laboratory Notices and Circulars etc. PROFS has been used extensively for Laboratory administration, particularly for document preparation, transmission and filing. Much material for this Report has been prepared using PROFS.
Following a decision that administrative database systems using INFO (a relational database package) should not reside on the PRIME B computer, they were transferred to the IBM environment under VM/CMS where INFO can also run. This involved extensive re-writing of control procedures. The RAL Library STATUS/INFO system has continued production and SPEED software has been installed to aid data input. Records are being added to the RGO Library database using the RAL data input system. A system to handle dosimetry records in the Laboratory has been completed. A system, FDS, has been set up, linked to PROFS but also available under VM/CMS, which maintains a database of financial information. The major software development project concerned the STATUS information retrieval system developed by AERE Harwell, which is being adapted to the VM/CMS environment. It is a particularly ambitious project involving use of modern techniques of intervirtual machine communication which can lead to distributed database handling. The work also involves interfaces to the IBM product ISPF and, of course, PROFS. Having STATUS available on the IBM system will allow databases currently mounted on RAL PRIMEs to be moved to the central computing system and made available to a wider community.
Participation in the database programme of the World Data Centre for Solar Terrestrial Physics was extended to the embryonic Geophysical Data Facility, initially with a catalogue and database for the NIMBUS satellites tape archive. The major development during the year has been a database for the recent successful AMPTE mission. This used generalised data management software (R-EXEC) developed by RAL computing staff. The move to unified software and user interfaces for all geophysical data services is already proving beneficial.
Hardcopy output previously destined for the FR80 is now produced on the Xerox 8700 laser printer. A system has also been developed for hardcopy output to be directed to the IBM 4250 electro-erosion printer giving a resolution of 600 dots/inch. Facilities for producing user titles on microfiche have been added to the service provided on the NCR 5330 microfiche recorder. A system has been developed which accepts metafiles in the format given in GKS annex E and produces output on the 8700 printer. Extra fonts and software have been installed in VM which provide for the typesetting of mathematical formulae on the 4250 and the RAL software which processes files to this device has been enhanced to allow output from the UNIX 'troff' system to be printed.
It was decided during the year to provide the NAG Graphical Supplement as a cohesive package of high-level graphics routines. These routines make obsolete those developed by RAL which cover some, but not all, of the functions available in the NAG system. Faults in the RAL routines as installed on the IBM systems have been cured during the year and a revised version of parts Band D of the RAL Graphics User's Guide has been prepared. A program which simplifies the plotting of experimental data (EZIPLOT) has been written; this allows casual users to produce any of the forms of data display provided by the high-level routines without detailed knowledge of the graphics system.
During the year, installation of GKS (the international standard graphics system, implemented jointly by RAL and ICL) has taken place on PRIME (under PRIMOS) and IBM (under CMS) computers. Initial releases of the PRIME and CMS systems were made in June and were accompanied by full user documentation. Extra device handlers have been developed and support for the Tektronix 4014, full Sigmex range and Benson plotter were under test at the end of the year. Throughout the year, a collaborative effort has been maintained with the Computer Board for Universities and Research Councils which is developing and distributing the RAL/ICL GKS system to its computers. System interfaces and device handlers developed by the Computer Board will add to the range developed by RAL.
The IBM 3032 system at RAL, which had been used as a development machine for SERC-wide administrative computing, was closed in July after systems running on that computer had been transferred to the new Joint Administrative Computing Service IBM computer at Swindon, a shared SERC/NERC machine.
EARN, together with its companion networks BITNET and NETNORTH, is a computer network between academic institutes and research centres for world-wide communications. It consists of a set of independent host nodes connected by leased telecommunication lines in an open, unmeshed topology. A central computer in each country provides international connectivity and some central information services. There are currently some 600 computers in the network. The Laboratory has recently developed a gateway between EARN and JANET, the Joint Academic Network. This allows the UK academic community to take advantage of the network with no additional equipment or software being required on site. EARN provides only file transfer and mail plus some messaging facilities which are provided by IBM computers. The gateway allows file transfer and mail facilities between the two networks with only minor restrictions. There is a continuous programme of development within EARN and, in particular, the European part of the network will migrate to internationally agreed standard protocols for data transmission. These standards will also be used within JANET and most other European countries within the next few years. In the meantime, EARN will provide a very valuable service allowing UK researchers to communicate with colleagues overseas.
A VAX 11/750 was upgraded by the addition of an extra megabyte of main storage, doubling its capacity, and a 414 megabyte fixed disk. This allowed the latest release of the VMS operating system and equivalent levels of networking software to be installed. With the increasing popularity of VAXs as local multi-user systems, RAL computing staff are now providing central support for nearly 50 machines. This support covers Digital's networking software as well as the coloured book software produced by UWIST and St Andrews.
The GIFT project is a collaboration aimed at producing a package of gateway software to run on a V AX computer which will allow file transfers between different but intersecting computer networks. It was originated by the European Committee for Future Accelerators in 1982 and is intended to link networks in Italy (INFNET), Switzerland (CERNET) and the UK (JANET). UK development has been undertaken by RAL in collaboration with Oxford University. This has involved the production of an interface between the file transfer software used to communicate with JANET and a network-independent file transfer system which runs in the gateway computer. The first stage of this development, which allows JANET users to initiate file transfers from their own computers to and from computers on CERNET and DECNET at CERN, was completed by the middle of the year. A production service was begun on 1 August on the JANET/CERNET gateway VAX at CERN. Further development and testing has continued on the second stage which allows similar file transfers to be initiated by CERNET and CERN DECNET users.
JNT is responsible for the establishment and introduction of common communications standards in the UK academic community. Its remit covers the activities funded by the Computer Board and the activities of the Research Councils themselves with the aim of creating a unified networking infrastructure. The programme of development and installation in support of the UK interim Coloured Book protocols has continued. Basic facilities for network access, file transfer and terminal communication are available on most commonly used machines, and support for electronic mail and job transfer exists for many systems. As an indication of scale, more than 350 electronic mail systems have now been registered with the Name Registration Scheme. Work has been completed this year on the definition of a further protocol, called the Simple Screen Management Protocol, which provides for improved full screen terminal working across the network. Some initial implementations of this protocol have been produced, and more are under way.
The growing interest in ETHERNET networks has led to the creation of recommendations for the set of protocols to be used with this technology. The JNT recommendations, which have been adopted by the Computer Board for use in university procurements, show the pattern expected in future years by adopting International Standards. The recommendations are Functional Standards; they list published international standards and state which of the options in these standards should be selected. They also define a mapping from the previously defined interim protocols. This is but one facet of a more general activity concerned with the move from the current interim protocols to the use of International Standards. A working group, drawn from university and Research Council networking experts, has been active throughout the year planning the transition to the new standards and determining the conversion facilities needed to prevent disruption of the service seen by network users. This group is due to publish its proposed strategy for comment.
This year has also seen a significant growth in JNT European involvement. It has taken a leading role in the initiative to establish the European Association of Academic Networks (RARE) to coordinate and encourage communication between institutions in Europe, and it has been involved in the CEN/CENELEC work aimed at the creation of a single set of functional standards for communication to open up a genuinely European market.
The Network Executive is responsible for the operation of the JANET network. Network rationalisation has continued throughout the year; all university and SERC sites and most NERC sites are now connected. Development has focused on consolidation, although significant increases in capacity have been made in the more heavily loaded trunk links. A programme of hardware upgrades has brought network switches to a common level capable of supporting the manufacturer's current control software. A phased transition to this software has taken place during the summer and early autumn and is now complete. This puts the network in a stronger position to adopt new products, allowing further enhancements. The major benefit so far of the move from SERC-produced software to manufacturer-supported software has been the availability of much more complete network management facilities. The logging of statistics allows much more information to be collected on the use of the network, and work has been in progress throughout the year on the processing, archiving and analysis of this data. The information available allows the Network Executive to monitor loading in different parts of the network, and trend analysis will allow links in need of upgrade to be identified before overload occurs. In the latter part of the year, planning for a major upgrade of the network began, to allow the support of more demanding applications. This work will continue throughout 1986, as will the existing rolling programme of rationalisation and enhancement.
