Virtualized HPC at Johns Hopkins Applied Physics Laboratory
Johns Hopkins University Applied Physics Laboratory (JHUAPL) has deployed a virtualized HPC solution on vSphere to run Monte Carlo simulations for the US Air Force. The idea was conceived and implemented by Edmond DeMattia at JHUAPL, and has been presented by Edmond and his colleague Michael Chinn at two VMworld conferences.
We now have a white paper available that describes this work. In a nutshell, the solution hinged on the fact that the Laboratory had a physical footprint constraint that restricted their ability to increase the throughput of their two Monte Carlo clusters, one running Linux and the other Windows. JHUAPL combined both clusters into a single, large virtualized cluster, creating what they dubbed a vGrid architecture. Overall throughput improved considerably, jobs complete faster, and each team often has access to many more physical resources than they did prior to centralization and virtualization.
See the paper for details.