Characterizing and Modeling Reliability of Declustered RAID for HPC Storage Systems

Declustered RAID aims to enhance the recovery performance of storage by shuffling data and parity blocks among all disks (including spares) in a RAID group. All drives in the RAID group participate in data reconstruction. The idea of declustered RAID is promising. However, the performance and reliability of declustered RAIDs in real-world High-Performance Computing (HPC) storage environments have not been thoroughly studied and well understood. With the popularity of the ZFS file system and software RAID used in production data centers, in this paper, we extensively evaluate declustered RAID with regard to the performance of RAID recovery comparing with ZFS RAIDZ. Furthermore, we formally analyze the reliability of declustered RAID in terms of the mean-time-to-data-loss (MTTDL) and discover that improved recovery performance leads to higher storage reliability compared with the traditional RAID.