1618 Tb SDDS storage BECOIN Proposal Upgrade operational

1618 Tb SDDS storage BE-CO-IN

Proposal • Upgrade operational LHC_DATA to 1. 1 Tbytes with the same operational level : – New faster technology (SAS disks instead of SATA) – RAID-1 • New service of ~16 Tb of READ-ONLY data containing ALL SDDS data : – Copied daily from LHC_DATA – No backup (but RAID 6 to secure the data) – Structured per year/month/day (TBD) 2 nd June 2009 LHC Beam Commissioning WG - P. Charrue 2

Proposal 2 nd June 2009 LHC Beam Commissioning WG - P. Charrue 3

Status • Both operational and read-only fileservers are now available • On the operational side, nothing changes, except CASTOR transfer which is now done from the new fileserver NFS 99 • NFS 99 has been loaded with ALL data from 2006, 2007, 2008 and 2009. 6 Tb (out of 18) are currently used. • See \cs-ccr-nfs 99data or /nfs/cs-ccr-nfs 99/data or \cs-ccr-samba 1slopsdataLHC_DATAOP_DATA or \cs-ccrsamba 1slopsdataSPS_DATAOP_DATA 2 nd June 2009 LHC Beam Commissioning WG - P. Charrue 4

Data flows 1. Operational applications write SDDS files on cs-ccr-nfs 4 2. Every night after 0 h 00 newly created files are copied from cs-ccr-nfs 4 to cs-ccr-nfs 99 3. Once per month, new files are compressed on cs-ccr-nfs 99 and sent to CASTOR 4. When needed, old files are deleted from csccr-nfs 4 to liberate some free space 2 nd June 2009 LHC Beam Commissioning WG - P. Charrue 5

Data Security • cs-ccr-nfs 4 is RAID-1: all files are copied 2 times on 2 different disks • cs-ccr-nfs 99 is RAID-6 and exported READ-ONLY. The server can cope with one disk failure • And finally all files are stored on CASTOR tapes • In case of major catastrophe : – If all cs-ccr-nfs 4 disks are corrupted, we can recover from the files copied on nfs 99 (but we may lose the production of the current day) – If all cs-ccr-nfs 99 disks are corrupted, we can recover from CASTOR tapes, but this will need about 3 weeks 2 nd June 2009 LHC Beam Commissioning WG - P. Charrue 6