摘要

In SSD-based All Flash Array (AFA), Redundant Array of Independent Disks (RAID) technologies are widely used to cope with drive failures and sector errors. However, the redundancy in disk level is either insufficient like RAID-5 or excessive like RAID-6. In order to combat the two types of errors, we combine the RAID-5 scheme with a Flash-aware Intra-disk Redundancy (FAIDR) scheme. For saving parity space, the ratio of redundancy in the FAIDR scheme grows with the increasing of sector errors. However, the parity handling overhead is significant. As NAND Flash blocks can't be updated before erase, we establish all FAIDR stripes on the physical page number of blocks regardless of the page state. Thus, the parity update operations in FAIDR are eliminated. Simulation results show that the proposed scheme has the ability to ensure the AFA's reliability in its all lifetime. It gains 41 percent storage capacity in comparison to RAID-6. And it only leads to very small performance loss compared with pure RAID-5.