Machine Room Planning, 6th Jan 2000

Machine Room Planning

Minutes of MRP meeting of: 06/01/2000. Present: T.Cass(chair), R.Minchin(secretary), D.Underhill, M.Vergari.

Computer room layout:
RPM has been asked to investigate producing a new floor plan. He reported that no new plan has been made since October 1997. Conversations with various people -Franck Jouberjean/IT/CIO (who produced previous floor plans), Mark Donszellmann/IT/IPT(Buildings at CERN), Marcel Mottier/ST/ISS (Building plans produced by ST) suggest that, despite being rather large, AutoCad provides the greatest ease of update. RPM has started a Tutorial and will investigate courses.

The idea taking shape is to provide a multi layered plan, available on the web but printable on A0, that would have:

1-3 layers showing the false floor and what is below it like holes to the vault, power distribution, network distributrion, etc,
1 layer showing different zones in the room such as PDP, PDP planned expansion(or foreseen obsolesence), NICE, Robots, Robot planned expansion, free space, etc,
1 layer showing the actual cabinets, racks and free-standing machines on the floor,
If possible, a blow-up of each rack or cabinet, showing the machines in it,
Information on a given machine - owner, responsible person, I/P address etc.

RPM could not predict when a plan would be available, but will try to produce at least an outline plan soon. ACTION: RPM

Understanding of different safety issues is increasing amongst the four people present.

We discussed whether or not we could accommodate the numbers of PC's required for LHC test bed applications - 400 PC's by end 2000 - we need a better understanding about the limitations of using Mezzanines for PC installations. For example, we don't know the floor capacity of the existing mezzanine (where there are currently conference rooms and the visitors' gallery). This subject needs to be followed up with A.Funken/ST. ACTION: TNT/DJU

Power off/UPS tests:
It is foreseeen to perform the Emergency Power Off tests on Saturday 29/01/2000 from 07:00-08:00.

Power will be switched to SPS pulsed power supply and will bypass the UPS.
Maintenance will also be done during this time on the automatic transfer between UPS and EDF.
During this period we would not be protected from micro power cuts.

It was also planned to perform a simulated discharge of the batteries of the UPS, module by module. This would take about three hours, during which time we would also not be protected. Given that the UPS batteries are coming to the end of their predicted lifetime and are due to be replaced it was felt that this test should not take place at this point in time. If it was felt worthwhile performing the discharge tests, they could be scheduled later when we could run on the normal power supply, although still bypassing the UPS. DJU will inform ST. ACTION:DJU.

We discussed whether or not systems should be shutdown before the test with particular emphasis on the difficulties that might be encoutered if REMEDY wasn't available for problem tracking. TNT will mail the necessary people and request that, if they are going to be shutdown, Data Base services are shutdown in the morning and not the night before.

Short term planning:

The move of the CE group machines of Antoint Thys will be followed up and should be completed by the end of January.

Long term planning:

Bat 186: Another meeting with ST is planned for early February, at whcih time we could also discuss some of the 513 issues (mezzanines, fire wall between Computer romm and barn, etc).
UPS battery life: Plans for replacement need to be circulated. (C5 have stated that UPS should provide sufficient protection until "good diesels" can take over)
Access Control: DJU will circulate prices for the different proposals.ACTION: DJU

Work Coordination in the room:
We discussed how to best coordinate the work done by Marcel Lasne (or others) in light of some comments that had been made suggesting some change need ed to be made. It was agreed that Marcel would meet with DJU, MV and perhaps Marc Collignon, after the Morning Meeting on Mondays in DJU's office to plan work for the week.

Obsolete equipment:
We agreed on the need to eliminate old equipment more rapidly. DJU is following up a process with SPL that allows for advertising equipment inside CERN, then, if nobody has expressed interest, advertising outside, and ultimately scrapping unwanted stuff.

Next meeting:
We discussed frequency of meetings and agreed that we needed to meet every two weeks initially, and monthly once things were running smoothly. Next meeting is Thursday 20/01/2000, 14:00.