THE JOHNS HOPKINS UNIVERSITY Next Generation Storage for the HLTCOE

The explosion of unstructured data in high performance computing presents a challenge for existing storage architecture and design. We present a combination of hardware and software which addresses the storage needs of our center’s compute cluster. We also demonstrate that at a constant total cost of ownership our proposed solution provides an order of magnitude better performance than the Johns Hopkins University’s GrayWulf cluster and is two orders of magnitude faster than the center’s existing storage array.