Abstract: Based on the intelligent early-warning, as self-recovered storage system is proposed to improve the reliability of Redundant Arrays of Inexpensive Disks(RAID) storage. This system takes advantage of machine learning methods to predict disk failures by collecting real-time Self-monitoring, Analysis and Reporting Technology(S.M.A.R.T) information on disks. A data-centric protection strategy using auto data migration is also implemented in the system to protect disks which predicts failure-prone. Analysis results show that this system can highly enhance the reliability of RAID storage, achieve access transparency for users and improve the availability of the system.

Key words: storage reliability, self-recovered storage, intelligent early-warning, storage system, disk failure

摘要： 为提高磁盘阵列存储的可靠性，提出一种基于智能预警的自恢复存储系统。在对磁盘S.M.A.R.T信息实时监测采集的基础上，利用机器学习方法对磁盘故障进行预测，对于预警磁盘采取以数据为中心的自修复迁移保护策略。分析结果表明，该系统能提高存储可靠性，数据迁移过程对用户透明，可用性较强。

关键词: 存储可靠性, 自恢复存储, 智能预警, 存储系统, 磁盘故障

CLC Number: 

• TP277