United States Department of Defense

Standard Queuing for HPC Centers Across the United States

Overview
HPCMP provides supercomputer services, high-speed network communications for defense laboratories and test centers to conduct research, development and testing across 10,000 nodes

Challenges

• Standardize shared resources

• Provide 4,000+ users a common workload management interface to all the supercomputing sites across the US

Solution

• Platform LSF HPC

• Platform LSF MultiCluster

Results

• Simplified process to submit work to a queue

• Standardized on common queuing user interface, workload manager, resource accounting, common plug-in scheduling

• Resource allocation decisions on consistent, accurate information

• Laid groundwork for multi-phase, virtual resource clustering project across multiple sites

"The High Performance Computing Modernization Program of the Department of Defense is committed to using proven, commercially supported, highly scalable grid computing technology in all of its shared computing centers. With the Platform solution, we are simplifying the process of users submitting work to queues through a common interface rather than multiple interfaces across multiple centers and machines. This standardized solution also gives us the benefits of common accounting and utilization reporting across all centers, so that we can base resource allocation decisions on consistent, accurate information."

Cray Henry, HPCMP Director