NERSCPowering Scientific Discovery Since 1974

Data Management

Data Transfer

STAR makes use of grid-based file transfers between PDSF and Brookhaven for two main purposes.  The first is to transfer input/output files for simulation jobs submitted from Brookhaven to/from PDSF.  The jobs come in through pdsfgrid.nersc.gov making use of the OSG software installation and the files are transferred through pdsfdtn1.nersc.gov to and from /project/projectdirs/osg.  The second purpose is for bulk replication of microDST's from Brookhaven.  In this case the final destination of the files is HPSS and the transfers are either done through pdsfdtn1.nersc.gov to one of STAR's elizas or through one of the NERSC Data Transfer Nodes.  Once the files are on disk at NERSC they are archived into HPSS in a separate step.

HPSS

STAR has historically been one of the largest users of NERSC HPSS and the two main things archived are the output from the production embedding jobs and the microDST's replicated from Brookhaven.  The data volumes for them is about 100TB/yr and 600TB/yr respectively.

Databases

The STAR calibrations database is mirrored from Brookhaven (RCF) to PDSF.  There are four database servers named pstardb[1-4].nersc.gov.  See their ganglia pages here.

STAR Scheduler

The STAR Scheduler is installed locally at PDSF.  For the complete documentation please see the STAR Scheduler homepage.  It is the main tool for users to use to analyze the datasets on the PDSF disks, and on many days thousands of jobs are submitted with the scheduler.  See the PDSF Scheduler Stats page to see its usage over the past three months.