SDSC Thread Graphic Issue 8, September 2006

RSS RSS Feed (What is this?)

User Services Director:
Anke Kamrath

Subhashini Sivagnanam

Graphics Designer:
Diana Diehl

Application Designer:
Fariba Fana

Allocations Corner

Data Storage Allocations and Project Extensions

—David Hart

The allocations staff at SDSC and TeraGrid sites have made it easier to request additional calendar time for projects proceeding more slowly than expected and to support projects with considerable data storage needs.

As a reminder, the next opportunity to submit MRAC proposals (requests up to 200,000 SUs) is from Sept. 15 - Oct. 13, 2006. The last MRAC and LRAC proposals were discussed and awarded Sept. 5-7.

Data Storage Allocations

SDSC has begun formally allocating several data storage resources via the same process by which researchers request compute time. Three such storage resources are now allocable: SDSC Collections Disk Space, SDSC Database and TeraGrid GPFS-WAN.

SDSC Collections Disk Space: This resource, currently more than 200 TB of disk managed by the SAM-QFS file system, are available for projects that need to share large data sets with geographically distributed collaborators or the broader research community. Allocations for this storage space should have a description of the collection, a well-defined (potential) user community, and a plan for making the collection accessible to others. This storage space is not appropriate for parallel I/O associated with a simulation on SDSC's HPC resources. SDSC DAC (start-up) proposals can request up to 1 TB (10,000 SUs) of Collections Disk Space.

SDSC Database: This resource provides projects with access to high-performance, enterprise-class database management systems (Oracle, DB2 and MySQL) running on dedicated SDSC resources. Up to 1 TB of total disk space may be requested via an SDSC DAC proposal; however, the core table space for a DAC request on this resource is limited to 30 GB. Once again, it is important for projects to have a description of the data to be stored, the need for high-performance database service, and any potential community sharing the data.

TeraGrid GPFS-WAN: The TeraGrid GPFS-WAN is a 200-TB parallel file system mounted on HPC compute resources at SDSC, NCSA and UC/ANL. A portion of this resource is being allocated to projects that have a need for long-term (12-month or more) storage of data sets. Data sets appropriate for GPFS-WAN allocations should be (a) tied to a long-term need for access to the data as part of computations on the relevant HPC resources, or (b) valuable to a broader community of research projects for their computational activities.

Projects with shorter-term needs for GPFS-WAN storage can request space via the request form. Up to 1 TB of GPFS-WAN long-term storage (10,000 SUs) will be available via TeraGrid DAC requests starting in October 2006.

Project Extensions

While allocation awards are, by default, awarded for 12-month periods, it is possible for PIs to ask for one extension to that time of up to six months. This has long been official policy, but how to submit such a request had been inconsistently handled among the sites.

That situation has now changed. To request an extension, you can now go to the POPS site. Once you login, you will see the option of submitting an "Extension" proposal type.

Extensions can be requested for just about any project proceeding more slowly than anticipated. Please note that no additional SUs can be requested as part of an extension.

Please contact David Hart, SDSC Allocations Coordinator, if you have any questions about submitting proposals, storage, or compute allocations.

Did you know ..?

that SDSC has limited the core file size to 32MB.
To make good use of the core file size it is recommneded using the MP_COREFILE_FORMAT environment variable (or its associated command-line flag -corefile_format) to set the format of corefiles to lightweight corefiles.
See Thread article Issue 4, February 2006 for more details.