RAL Tier1 Resources Review 20110302
Contents
- 1 Agenda
- 1.1 Overview of purpose of meeting, review agenda and meeting mechanics (Glenn Patrick)
- 1.2 Tier-1 deployed capacity and near term projections (Matt Hodges)
- 1.3 Experiment resource usage (Andrew Lahiff)
- 1.4 Experiment requests and proposed allocations (Glenn Patrick)
- 1.5 Summary of MoU commitments
- 1.6 Financial situation and longer term capacity planning (Dave Britton)
- 2 Attendees
- 3 Notes
- 4 Additional documents
Agenda
Overview of purpose of meeting, review agenda and meeting mechanics (Glenn Patrick)
Tier-1 deployed capacity and near term projections (Matt Hodges)
Experiment resource usage (Andrew Lahiff)
Experiment requests and proposed allocations (Glenn Patrick)
Summary of MoU commitments
Financial situation and longer term capacity planning (Dave Britton)
Attendees
Dave Britton, Roger Jones, Chris Brew, Jens Jenson, Matt Hodges, Dave Sankey, Andrew Lahiff, Andrew Sansum, Shaun de Witt, Glenn Patrick, Brian Davies, Stefania Ricciardi, Raja Nandakumar
Notes
Tier-1 deployed capacity and near term projections
The majority of the 2010 CPU procurements will be commissioned during March, with a small remainder in April. SL06 will be decommissioned in late 2011 or early 2012. There are no problems meeting the 2011 WLCG CPU pledges. The 2010 disk procurements are likely to be commissioned in Apri, and SL08 is anticipated to go back into production in June. There is a slight reduction in disk capacity during 2011 as the CV05 and V06 diskservers are decommissioned. The 2011 WLCG disk pledges can easily be met, with a large amount of headroom. The 2011 WLCG tape pledges cannot be met until the T10KC tapes go into production. This is currently expected to take place in October. Originally T10KC was expected to go in production in June, this has been delayed due to, for example, a CASTOR upgrade required in order to handle 5 TB tapes.
Experiment resource usage
There has been no cause for alarm in the CPU, disk and tape usage over the past 6 months. Tape usage over the past year has been almost linear, so unless the VOs start writing dramatically more data to tape we don't expect any problems despite not being able to meet the WLCG tape pledges until October.
Experiment requests and proposed allocations
No allocations have been made for 2012 due to the uncertainly of LHC experiment requests. There is an overhead of over 2 PB of disk. Possible uses of this overhead need to be discussed.
Summary of MoU commitments
The WLCG CPU pledges have been met over the past 6 months, but the total CPU capacity hasn't been high enough in order to give 10% to the non-LHC VOs. Over the past month the disk pledge has been met for all LHC VOs, but there have been some issues not meeting them recently due to diskservers in intervention. There was a suggestion that each VO being slightly over-allocated may help to avoid this in the future.
Matt H will allocate enough disk to each VO so that the CV05/V06 decommissioning won't require any new deployments later in the year.
Financial situation and longer term capacity planning
There is currently uncertainty with the 2012 WLCG pledges. New experiment resource requests are expected in April. The Tier 1 is expected to be in good shape financially over the next few years.
