Reliability Analysis of Declustered- parity RAID Focusing on Uncorrectable Bit Errors

Abstract

UNcorrectable Bit Error(UNBEs) have been ignored or have not studied in detail in existing analysis of the reliability of RAID, We present an analytic model to study the effect of UNBEs on the reliability of declustered-parity RAID. The Mean Time To Data Loss (MTTDL) of the RAID is calculated by tasking into account both DB data loss and the data loss caused by double independent disk failures (we call this DD data loss). On the basis of our numerical analysis, we discuss how such MTTDL depends on the number of units in a perity stripe and the rebuild time of a failed disk.