The National Energy Research Scientific Computing Center (NERSC) in the US has upgraded its ‘scratch’ storage for temporary scientific data to a single site-wide system with four petabytes of storage built on DataDirect Networks (DDN) Storage Fusion Architecture (SFA) high-performance storage engine. By selecting a unified scratch file system, NERSC was able to build to the precise levels of performance and capacity required while optimising the cost of configuration at every step. This continues NERSC’s strategy to move away from distributed storage ‘islands’ in favour of ‘global’ storage.
By embedding file systems in the storage controller, DDN’s converged infrastructure approach enabled further optimisation by removing latency and a significant number of network connections and servers. As a result, NERSC was able to meet performance and capacity requirements at 30 per cent lower cost of storage than implementing local storage and file systems for each compute platform. The ability to write temporary data to a central repository for further analysis has also enabled NERSC to reduce its local ‘scratch’ storage costs by more than 50 per cent, while hundreds of thousands of dollars in infrastructure costs were saved by eliminating the need for additional servers, cabling, network switches and adapters.
In terms of performance, NERSC has been able to achieve 80 GBps with the fewest number of systems, which also has resulted in reduced administrative overhead and lower data centre costs.