Sustained System Performance (SSP) Test Job Execution
BC Project: FY10-04
Date of Policy: 13 Jun 2011
Last Updated: 16 Aug 2018 (see Revision Log)
Notice: The Sustained System Performance (SSP) test jobs are being revised to allow easier implementation and maintainability of the allocated HPC systems at all DoD Supercomputing Resource Centers (DSRCs). This policy is currently suspended until a new collection of supported test jobs has been identified and implemented.
The purpose of this policy is to mitigate noticeable degradation of system performance after scheduled or unscheduled events.
All DSRCs are required to submit SSP test jobs for execution after specified outages on their allocated systems. The Benchmarking Team will be responsible for providing the DSRCs with the appropriate test jobs for each system’s architecture. If a particular SSP suite is not made available to any DSRC, that DSRC is not obligated to run the SSP test jobs. These jobs will be the same as those currently run by the Benchmarking Team on a quarterly basis.
SSP test jobs must be run after scheduled or unscheduled events which include, but are not limited to, the following: operating system upgrades, security patches, power outages, or other events at the discretion of the DSRCs.
After a system has been placed back into operation, SSP test jobs should be submitted by DSRC personnel who will monitor the jobs to ensure their completion. The Benchmarking Team will review the results and provide guidance to the DSRC as necessary.
In addition to executing the provided jobs for compliance with this policy, the DSRCs are strongly encouraged to perform their own verification and validation tests to ensure that the overall functionality of the system is satisfactory before it is returned to service.
|16 Aug 2018||BC Team Audit - Reworded the "Notice"|
|26 May 2016||BC Team Audit|
|25 Apr 2014||BC Team Audit - Notice on "Compliance is on hold" included as a header in the policy|
|25 Apr 2012||BC Team Audit|