In this video from the 2014 HPCAC Stanford HPC & Exascale Conference, John Fragalla from Xyratex presents: Improving Lustre OST Performance with ClusterStor GridRAID.
HPC storage solutions relying on RAID 6 data protection experience Lustre OST degradations and reduced filesystem performance, including potential data loss, during drive failures, rebuilds, and recoveries. ClusterStor GridRAID delivers up to 400% faster drive rebuild times, mitigates drive failures, increases system resiliency, reduces data loss, and maintains peak HPC application performance.”
Download the slides (PDF) * See more talks at the Stanford HPC Conference Video Gallery
In related news, OpenSFS will host the 12th annual Lustre User Group conference April 8-10 in Miami, Florida.
This is no different than Declustered RAID in IBM GSS.
I believe the roots of this parity declustered RAID technology go all the way back to this paper by Mark Holland and Garth Gibson: http://www.pdl.cmu.edu/ftp/Declustering/ASPLOS.pdf