UCL HEP Computing Status HEPSYSMAN RAL 2005 04

  • Slides: 9
Download presentation
UCL HEP Computing Status HEPSYSMAN, RAL, 2005 -04 -27

UCL HEP Computing Status HEPSYSMAN, RAL, 2005 -04 -27

Computers ● Desktop PCs: ~50, from 733 MHz to 3. 2 GHz ● Various

Computers ● Desktop PCs: ~50, from 733 MHz to 3. 2 GHz ● Various laptops, inc. a few i. Books ● Group batch farm: 10 dual 2. 4 GHz Dell rackmount ● CDF batch farm: 10 dual 2. 4 GHz Dell rackmount ● ATLAS trigger testbed: 6 rackmount machines ● Mail and web servers ● LCG front-end nodes (CE, SE, MON) ● Windows Terminal Server ● Various dedicated and development machines

Operating Systems ● Desktops, farms, servers almost all SLC 3 – ● recently completed

Operating Systems ● Desktops, farms, servers almost all SLC 3 – ● recently completed changeover from RH 7. 3 Windows 2000 Terminal Server – some problems with SAMBA ● Windows machines for Auto. CAD, hardware control ● Laptops a mix of Linux, Windows, OSX

LCG Front-End Machines ● Dedicated service nodes: CE, SE, MON ● Jobs go to

LCG Front-End Machines ● Dedicated service nodes: CE, SE, MON ● Jobs go to HEP batch farm (normally) ● Support for separate CE and batch server – not provided as part of standard LCG – relies on recipes from third parties – is getting easier – well documented for LCG 2. 4. 0 by Steve Traylen

Central Computing Cluster ● UCL SRIF-funded facility ● 96 dual 2. 8 GHz P

Central Computing Cluster ● UCL SRIF-funded facility ● 96 dual 2. 8 GHz P 4 nodes ● Half dedicated to LCG ● Other half uses Sun Grid Engine (non-HEP use) ● Managed by Information Systems ● Hope to integrate HEP and non-HEP parts – use SGE info provider from Le. SC – but need to change again for g. Lite?

Storage ● User file server with ~300 GB ● One 1. 2 TB IDE

Storage ● User file server with ~300 GB ● One 1. 2 TB IDE RAID for data ● One 1. 2 TB IDE RAID for backup ● One 3. 2 TB RAID for data/backup ● Various RAIDS bought by MINOS and CDF ● Tape drive for backup

File backup ● ● ● Up to now: – backup selected areas to disk

File backup ● ● ● Up to now: – backup selected areas to disk (tar) – secondary backup to tape Problem: – tape solution expensive – need to spend more money to use with SLC Solution: – RLBackup – Currently keeping old disk backup + RLBackup

Changes ● Bob Cranfield retiring later this year ● Ben Waugh to be group

Changes ● Bob Cranfield retiring later this year ● Ben Waugh to be group “computing coordinator” ● Gianfranco Sciacca started recently ● Machine room moving to new building (early 2006? ) ● Moving behind campus firewall

Issues ● Desktops vs laptops ● Management of laptops – OS choice – support

Issues ● Desktops vs laptops ● Management of laptops – OS choice – support – updates ● Remote administration of servers and farms ● Management and support using fractions of people's time