CERN home pageCERN home pageDocuments by ReferenceDocuments by ReferenceCNLsCNLsYear 2000Year 2000Help, Info about this page

Contents

Editorial Information
Editorial
If you need help
What our Users Have to Say ...
Announcements Physics Computing Desktop Computing Internet Services and Network Scientific Applications and Software Engineering The Learning Zone User Documentation
Previous:HPSS Stoppage During week of 8 January 2001 and 2001 Data
Next:RISC Decommissioning
 (See printing version)



IT/PDP Plans for 2001

Tim Smith , IT/PDP


Abstract

Evolution of PDP services in 2001.


In 2000 we have spent much energy on the aligning the configurations of the numerous individual experiment facilities, such that we could start to merge them into larger shared facilities: for example LHCb and L3C are now part of LXBATCH. This has proved very successful for both the public batch facility LXBATCH and the shared allocatable resource EFF. During 2001 we will be seeking to extend this concept by phasing out the EFF and replacing it by an even larger TESTBED facility and by merging more of the dedicated experiment clusters into LXPLUS/LXBATCH. This has only been possible due to the across the board uptake of PC technology. Along these lines, we have proposed a RISC decommissioning schedule to FOCUS (see the separate article) and will be continuing to replace aging RISC equipment with PCs wherever appropriate. The only exception to this is the proposal to create a new, shared Solaris facility SUNDEV for code development and verification, in line with the long-term strategy to consolidate onto just Linux with Solaris as the 2nd platform.

PC technology has also taken over in the area of Data Management in 2000. All tape servers are now PCs as are a growing number of disk servers. Following successful production usage of EIDE disk servers for L3C and ALEPH, we have proposed to buy EIDE disk servers instead of SCSI disks for RISC machines in 2001. They already run in mirrored configurations and we are now testing RAID-5 configurations, which open up very attractive possibilities for easing operations. CASTOR, the in-house software replacement for SHIFT, has been successfully deployed on several servers, giving us confidence that we can now base central services on it during 2001. To this end the HPSS service, which has been deployed for CDR services in 2000 will be cut back in 2001, and be replaced by a central CASTOR infrastructure for CDR. Finally to enable more efficient use of tapes and tape infrastructure, and to enable more transparent migration to new media, we propose to move away from the mode where users deal with individual tapes, to a managed storage mode. This will be announced separately when ready for deployment.


About the author(s): T. Smith is the Infrastructure and Services Section Leader in the PDP group.


For matters related to this article please contact the author.
Cnl.Editor@cern.ch


CERN-CNL-2000-003
Vol. XXXV, issue no 3


Last Updated on Mon Dec 18 13:51:01 GMT+03:30 2000.
Copyright © CERN 2000 -- European Organization for Nuclear Research