IBM SPECTRUM SCALE FOR GPFS
NVMe-oF Storage for IBM Spectrum Scale GPFS
Parallel File systems like GPFS require shared storage and until now its been deployed in shared-disk or shared-nothing distributed parallel nodes. Recently, the only way to take advantage of NVMe’s low latency and parallelism with Spectrum Scale GPFS was to use either:
- Legacy All-Flash Arrays limited to a pair of active-active controllers with NVMe drives.
- Software Defined Storage (SDS) that adds processing overhead and management challenges to every server in the cluster.
However, due to recent technology advancements Pavilion now delivers shared storage with orders of magnitude more throughput for parallel file systems like GPFS. Whilst its loaded with the manageability features you expect from a Storage Area Network (SAN), it comes with 10x the throughput, important data management features and no single point of failure.