摘要
冗余磁盘阵列虽然引入了容错机制使得磁盘阵列的数据可靠性得到了很大的提高，但同时也引起性能下降。而且随着磁盘设备数量的增加，磁盘失效的概率将明显增大，当单个磁盘失效后，虽然此时磁盘阵列数据并未丢失，且仍能服务于系统的请求，但此时磁盘阵列是带“病”工作，处于一种降级模式，本文对冗余磁盘阵列ＲＡＩＤ５进行了队列建模和仿真计算，提出了性能损失率的概念，并作为评价磁盘阵列性能损失的衡量指标。计算结果分析表明，ＲＡＩＤ５在降级模式下的性能将为整个阵列性能的瓶颈。
The reliability of redundant arrays of inexpensive disks(RAID) have improved dramatically by redundant disk, but the performance of RAID will decrease, and the probability of the disk to fail becomes larger with the array size increasing. When a disk failed, the RAID system did not loss any data of the host, and continued serving for the host. But the RAID would be working in a degrade mode. This paper submits the queueing model of RAID5, and proposes a new concept: Performance loss rate is as a parameter to evaluate the RAID's performance. The results show that the performance of RAID5 in the degrade mode is the bottleneck of the disk arrays.
出处
《计算机工程与设计》
CSCD
北大核心
1995年第3期9-19,8,共12页
Computer Engineering and Design
基金
国防科工委基金
