Dear David & all, Plasma has 3--4, rather old, Beowulf type machines down there. We probably only need 1, well looked after machine. Most "production runs" are done on cx1 (cx2) or other HPC facilities that we get time on. Plasma are trying to get aircon put into the small darkroom next to my office, so that we can have 1 modest machine in there (i.e. 1 rack with ~16 nodes/blades). This would be a designated "PhD training machine" where they can safely be let loose, and also have a more transparent system for parallel code development & optimization (i.e. log onto any node, no queuing system, etc.). If this room does get created, then Plasma will not need use of H147 for computing. (That's the consensus of the academics here.) Richard - if (when?) we have to move out of H147, will SPAT still need our rack? Best wishes, Robert PS: David, will you be going to the examiners meetings ;-) ? (What will happen to you there?!) Actually, I've been tempted to contribute to the debate by mentioning that 1) I had to rearrange my forthcoming wedding to fit in with the meetings. (Our preferred date was 25th June, but we decided it would be rubbish to then have to wait a week to go on honeymoon because of these meetings; so we moved it to September!) 2) I have been unable to attend the big EPS Plasma conference for 7 years (i.e. since it took place at IC) since it always falls in the week of the meetings. I suspect, this has damaged (or at least not helped) my research career, as I have missed out on numerous opportunities to give invited talks there. So my question is: Just how important is the examiners meeting? Is expecting everyone to attend, given situations like the above reasonable? I'd definitely like to clarification! (Maybe I should be regularly asking for leave from these meeting, but I've always been under the impression this is frowned upon.) On 9 Jun 2011, at 10:53, Bantges, Richard J wrote:
Dear David (et al.),
SPAT currently hosts 3 rackmount NAS boxes in H147 (with UPS's), with the prospect of an additional 2 arriving shortly.
I would be extremely unhappy about moving these systems to Mech Eng as we backup some of these on systems that are already housed in Mech Eng. A while ago we were told that after the IT disaster, following a fire, at Southampton University, that we needed to provide offsite backup of our core data. The feeling was that Mech Eng is sufficiently far away from Huxley, that this was considered appropriately "offsite". I'm aware that ICT can provide a service to back up data offsite (e.g. data from Mech Eng is backed up at the Hammersmith Campus etc.), but these are services which have monthly / annual service charges.
That said, we (SPAT), don't need a huge amount of space, but would need a rack with space for at least 5x 4U NAS boxes and a 1U UPS. These cannot be housed in an office as they're simply far too noisy.
Rich
-----Original Message----- From: physics-departmental-computing-bounces@imperial.ac.uk [mailto:physics-departmental-computing-bounces@imperial.ac.uk] On Behalf Of David Colling Sent: 09 June 2011 10:29 To: physics-departmental-computing Subject: [Physics-Departmental-Computing] Hosting equipment in the Huxley machine room
Dear All,
I know that this was discussed at the last departmental computing meeting but things seem to have moved on. So I need a list of who hosts machines in the machine room in Huxley and what arrangements they would like if we are turfed out to make room for a big laser. In particular is there anybody who would be unhappy simply to move their machines over to Mech Eng and host them in ICT's machine rooms? If so why? I am not completely sure how quickly I need this information but the email from Jo implied sooner rather than later ...
Best, david
_______________________________________________ Physics-Departmental-Computing mailing list Physics-Departmental-Computing@imperial.ac.uk https://mailman.ic.ac.uk/mailman/listinfo/physics-departmental-computing
_______________________________________________ Physics-Departmental-Computing mailing list Physics-Departmental-Computing@imperial.ac.uk https://mailman.ic.ac.uk/mailman/listinfo/physics-departmental-computing
-- ============================================================================= Dr Robert J. Kingham Plasma Physics Group, Tel: +44 (0)207-5947637 Room 724, Blackett Laboratory, Fax: +44 (0)207-5947658 Imperial College London, E-mail: rj.kingham@imperial.ac.uk London SW7 2AZ, UK. Web: Professional Web Page =============================================================================