Q1 How much will it cost to use the CDR service

The use of the CDR service itself is free of charge. Experiments will, however, have to pay for the use of the magnetic tapes on which their data will eventually be stored. For the present, data storage on magnetic tape is charged at the rate of ~2 CHF/GB. Only the total volume of data is charged for, not the actual number of tapes used. The cost is calculated at the end of the year. Experiments whose data rates are low, typically <3-5 MB/sec, will be able to use the CDR public servers to write their data to CASTOR. Experiments anticipating data rates > 5MB/sec will need to purchase disk server(s) to be used as CASTOR and/or analysis servers. The cost of this is borne by the experiment.

Q2 What are the system requirements for using CDR

AFS is not necessary in order to run CDR. Where AFS is installed, it is recommended to have local copies of the executables on disk rather than running executables out of AFS. The same applies to dynamically shared libraries. Local copies make the CDR independent of irregularities in the AFS service.

CDR runs under its own dedicated account. The account name should be registered as an AFS account and it is strongly recommended that a local entry in /etc/passwd be created with same uid/gid combination.

No. However, experiments wishing to run during this period should contact cdr.support@cern.ch to see if something can be arranged.

Contact cdr.support@cern.ch so that details can be finalised.

The main activity of CDR is the remote file I/O copy process which reads a local file and writes to the castor staging server. The rfcp program is multithreated but for all testbeams, only one copy of the process runs at a time.

In addition to the rfcp process, there are several scripts which run periodically, typically every few minutes. These scripts are concerned with logging and file cleanup and do not create any significant load.

Yes. In general CDR only copies files from the DAQ to tape storage in CASTOR. Modified versions of the CDR scripts have been developed to handle Objectivity databases but these are specific for individual experiments.