It is quite clear from the paper to be put before Council at their next meeting that there is little likelihood of the current Atlas Computer Laboratory continuing to exist for more than a few years. The main reasons for this are:
I think the staff of ACL will be living in cloud cuckoo-land if, with these constraints, they think there is still a chance of a separate ACL existing on the Chilton site. Due to the successful lobbying of the Nuclear Physics community, powerful computing facilities exist at both Daresbury and Rutherford while ACL has a system which can only rival them in flexibility.
One constraint, which the Council was reluctant to admit initially, is the need for an ACL function. University research workers recognise the merit of having large computer facilities where they can get guaranteed amounts of computing time which suits their needs. Also, there is a need for ancillary services to supplement the basic computing power. With the advent of communications networks and RJE stations, the user has the capability of accessing these resources in a manner far superior to the one available when ACL was first set up. The Regional Centres have not fulfilled this ACL function and it seems likely that SRC will need to support it for a considerable time in the future.
The main parts of the proposal put forward by London Office are:
The first two parts of these proposals are quite sensible and are likely to have sufficient backing to be accepted. Moving all Nuclear Physics work to Rutherford, and constraining it to the Rutherford budget, should ensure that capital is available for other projects. This should release at least £2m a year. This, together with the closure of ACL at Chilton, should allow £3m or £4m to be available.
Assuming that we start with the ground rules put down in the Introduction, together with (1) and (2) above, the proposal (3) is a possibility. There will be a continuing need for computer resources on the Daresbury site. However, some spare computing power should be available for the ACL function. The 370/165 is a powerful machine which could be upgraded to a 370/168 at a cost of less than £lm.
There are, however, a number of major problems with this proposal which make it unworkable. Most of the staff at ACL are well aware of these and I will do no more than outline them here:
From the previous two sections it is evident that, if Council keeps within its own ground rules and constraints, there is little room for manoeuvre. Even so, a more logical solution can be found. I feel the proposal outlined in the paper has a number of conclusions which are not arrived at by logic - instead, they are motivated by fear of the Nuclear Physics community and the need to do the right thing by that good and faithful work-horse, ACL. In addition, it is evident that a number of purely computer problems have been overlooked. For example, the differences between the 360/195 at Rutherford and the 370/165 at Daresbury are nearly as great as those between the 360/195 and the 1906A. The cost and feasibility of running a substantial load on the 360/195 from Daresbury has not been examined.
Any proposal must satisfy the conditions that a national ACL facility is provided and that at least 20% of the 360/195 should be used for this purpose. With these constraints, the only solution possible is that, assuming ACL closes, the Rutherford must take over at least part of the ACL function. This seems to have been resisted in the paper for reasons which are not plain. In the case of the Nuclear Physics work, the paper specifically states that, although this work would normally be done at Rutherford, there may be occasions - perhaps due to geographical location - when Daresbury would be more appropriate. Similarly, for geographical and computing reasons, it is likely that Rutherford will always need to do some of the work associated with the ACL function. Once this has been accepted, the correct course to follow becomes quite obvious:
Item (3) is presumably the most controversial. It does, however, make a great deal of sense. The current ACL users are predominantly in the South and this is partly due to geographical considerations. There is a need for users to visit the facility from time to time. Also, telephone charges are such that teletype links and, to a lesser extent RJE stations, should be as close as possible to the central facility.
If two ACL functions were set up, it would be sensible to have them complementary. The large Nuclear Physics load on the Rutherford machine probably means that the facility will be less flexible than the Daresbury one. It could be that one Laboratory should concentrate more on graphics than the other. Large data processing tasks should be concentrated, perhaps, on one facility. To some extent, this situation already exists on the Chilton site. Work is placed on the 1906A or 360/195 depending on its characteristics. Similarly, allocation of time by SRC committees, would take note of the form of the computing requirement, together with the load on the two Laboratories.
One major asset of this proposal is that the traumatic move of ACL to Daresbury over a short timescale would no longer be necessary. The second asset would be that a suitable use would be made of the ACL buildings.
The major reason for delay in the Council plan is that the current 1906A facilities at ACL have to be run down and the build-up of Daresbury is dependent on this. This new proposal would allow a much quicker build-up of the Daresbury facility. The major actions would be the setting up of the two computer facilities at Rutherford and Daresbury. The current Divisions are probably inappropriate as both are too well connected with site automation and electronics respectively. It is possible that both the North and South facilities could be given separate Directors from the site Director. It is unlikely that this would work and I think it is more sensible to have the Computing Division as a strong Division in each Laboratory with responsibilities to the University community recognised. This may only need a Council decision indicating how much of the total computing resource is available for on-site work.
There seems no reason why this should not go ahead with the utmost speed. The important thing is that the Head of both Divisions be appointed and the basic structure organised. In the case of the Rutherford Division, this is likely to be formed mainly from the current C & A Division staff, but should also include ACL personnel. The Daresbury facility is likely to require more of the ACL staff as it is here that the major work will be required.
I would imagine that ACL staff would gradually move to one or other Laboratory as the build-up takes place. At some specific date, the remainder of the ACL facility at Chilton would be absorbed into the Rutherford Division. To make it quite obvious that this is not the absorbing of ACL by the Rutherford C & A Division, it would be sensible for the Headquarters of the new Division to be located in the ACL building. Any new Rutherford machine would, most likely, be placed in the old Atlas building. It would eventually be the Rutherford Laboratory's responsibility to provide a service on the 1906A until it was removed.
At Daresbury, there would be a need to get the ACL function off the ground. This would require considerable help from current ACL staff and a significant number of staff would move to Daresbury in the next 2 years. The aim here should be to build up the Northern Laboratory so that the computing facility was of the same calibre as that at the Southern site. Due to the quite long period before the Nuclear Physics work is absorbed at Rutherford, it is essential that money be spent building up the Daresbury facility. For example:
I think the suggestions made above are more likely to be successful than the London Office proposal. However, in any decision, it must be made clear that the Daresbury site will receive a large increase in computer resources. It must be made clear that both North and South centres are primarily Council facilities, rather than Board facilities. This could perhaps be emphasised in the case of the Rutherford Laboratory by a change of name to Chilton Laboratory.
The proposals would be much more sensible as far as ACL is concerned. The expertise available in the Laboratory is more likely to remain under this proposal. Staff are likely to join both the North and South Laboratories and are likely to affect the character of these places. Consequently, the assets of ACL will not be squandered away.
