Disk array system, data writing method thereof, and fault recovering method

摘要: In a disk array system for RAID (level 5) to improve process performance by dispersing data, the overhead during data writing operation is reduced. Even when write are used rewrite #1 #4, which have already been written at address SADR 2, SADR1 and 3 within drives as belonging mutually different parity groups, these considered new then in parallel into, instance, an empty region drive 4. The updated old or not read out. invalidation flags registered into conversion table. subsequent readout carried out from newly region. When all of original group invalidated, holding this valid group, partially replaced proper timing. other words, acquired produce will be stored above operation, it assumed that length transferred host unit constant, one has constant length.