The current CDR system is a software tool for the migration of
raw data to CASTOR,
the CERN developed mass storage system (MSS). The CASTOR system
manages data files that are stored on a variety of cartridge tape
devices as well as the migration of data to and from data storage
pools on disk servers. Storage pools are groups of UNIX filesystems
managed by a CASTOR stage daemon and it is in these storage pools
that data files are processed by physics analysis programs. A central
database holds the names of all files in the CASTOR name space.
To facilitate this process, dedicated CDR-CASTOR public diskservers
have been installed.
The most common use of the CDR system is for LHC testbeam data.
However, more demanding experiments like NA48, COMPASS and HARP have
used the CDR system to transfer data at up to 35MB/sec using multiple
data streams. Variants of the basic CDR service are also used to
manage data produced in large Monte Carlo productions and mock data
challenges for LHC experiments.
Central Data Recording at CERN was pioneered by Bernd Panzer-Steindel
in the context of CERN's participation in the Esprit project. The
FOCUS comittee approved the service in March 1996 and now
all CERN experiments use it as their normal method of data
recording.
|