INFN – CNAF is the main Italian LHC data centre and one of the few
primary level (Tier-1s) LHC data centers in the world. LHC experiments
will produce up to tens of petabytes of data which need to be stored at
the various LHC computing centers for reconstruction and analysis.
Different data management policies are provided, identified by 3 distinct
Storage Classes:
• Disk0-Tape1 (D0T1): data migrated to tape and deleted from disk when
the staging area is full. Disk and tape space is managed by the system
CASTOR (testing GPFS/TSM/StoRM)
• Disk1-Tape0 (D1T0): data always available on disk, never migrated to
tape, never deleted by the system. Space management is demanded to
the experiment GPFS/StoRM
• Disk1-Tape1 (D1T1): large buffer on disk with a tape back-end. No
garbage collector, space is managed by the experiment.
Moreover, an Oracle clustered database infrastructure is deployed for
relational data CASTOR (moving to GPFS/TSM/StoRM)
Storage
- Details
- Written by Anna Karen
- Hits: 624