摘要： 随着磁盘容量的不断增大，磁盘读写出错的概率也相应增加。为缩短磁盘阵列在磁盘出错后所需的恢复时间、降低磁盘阵列重构的频度，提出一种磁盘阵列的多级恢复系统MARS。该系统根据磁盘的不同健康状态调用盘内坏扇区映射MARS-1，盘间数据迁移MARS-2和磁盘重构MARS3恢复磁盘阵列。实验结果表明，多级恢复系统能有效减少磁盘阵列恢复的时间，降低磁盘阵列的重构次数，减少对系统I/O性能的影响。

关键词: 磁盘阵列, 数据可用性, 多级恢复, SPC trace测试

Abstract: As disk capacity increases, the probability of partial disk failure increases. This paper presents a Multilevel Array’s Recovery System (MARS)to save recovering time when recover a degraded array and reduce the rebuild frequency. The MARS system will choose bad-sector- remapping(MARS-1), or data-migrating-between-disks subsystems(MARS2), or Redundant Array of Independent Disk(RAID) rebuild to recover RAIDs according to different disk health status. Experiment results shows that the MARS system can save recovering time and reduce the rebuilding frequency obviously. Besides, the I/O performance of the RAID may not drop during the recovering.

Key words: Redundant Array of Independent Disk(RAID), data availability, multilevel recovery, SPC trace test

中图分类号: 

• TP333