主动容错云存储系统的故障预警调度策略  

Failure pre-warning scheduling strategy for cloud storages systems with proactive fault tolerance

作　　者：李国[1] 曾婷婷 曹卫东[1] 李静[1] LI Guo;ZENG Ting-ting;CAO Wei-dong;LI Jing(School of Computer Science and Technology,Civil Aviation University of China,Tianjin 300300,China)

机构地区：[1]中国民航大学计算机科学与技术学院,天津300300

出　　处：《计算机工程与设计》2022年第12期3388-3395,共8页Computer Engineering and Design

基　　金：国家自然科学基金民航联合基金项目(U1833114);中国民航大学中央高校基金项目(3122019122)。

摘　　要：针对主动容错云存储系统中现有的预警调度机制存在等待时间长、易发生数据丢失等问题,在主动预测硬盘故障的基础上,设计一个预警调度模型并提出一种融入动态优先级的预警调度算法DPS(dynamic priority-incorporated scheduling)。利用决策树算法进行硬盘故障预测,根据预测结果分析硬盘健康度,综合考虑系统中RAID-5和RAID-6校验组的分布,利用DPS确定预警事件的优先级并分配带宽资源。仿真结果表明,在系统带宽资源的限制条件下,相比默认的调度策略,该策略在不同的RAID重构时间和操作人员反应时间下均能降低云存储系统中的数据丢失次数。Aiming at the problems of long waiting time and easy data loss in the cloud storages systems with proactive fault tole-rance,a pre-warning scheduling model was designed and a dynamic priority-incorporated scheduling algorithm(DPS)was pre-sented.The decision tree was used to predict disk failure and disk health degree was analyzed according to the prediction results.The distribution of RAID-5 and RAID-6 parity groups was also considered comprehensively in the system.The DPS was used to determine the priority of pre-warning events and allocate bandwidth resources.The simulation experiment results show that comparing with the default strategy,under the limitation of system bandwidth resources,the proposed strategy can reduce the times of data loss events for different rebuilding time of RAID and the response time of operator in the cloud storage system.

关 键 词：云存储系统 硬盘故障 决策树 动态优先级 调度 预警 RAID 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]