HR-RAID: Double failure tolerant RAID scheme with High Reliability

摘要

Existing parity RAID is a redundancy code capable of correcting any single self-identifying failure. Recently, Big Disk array appeared. Considering system crash, uncorrectable bit error and correlated disk failures, multiple failure chances are increased. Thus we propose multiple fault tolerant RAID scheme with two-dimensional redundant information (HR RAID). We show that HR-RAID has highest reliability and improved small write throughput in comparison with RAID level 6.