摘要
本文比较分析了纠单错阵列与纠双错盘阵列的可靠性；指出若只考虑盘一级可靠性，纠双错阵列平均无故障时间（ＭＴＴＦ）是纠单错阵列的几千倍；介绍了一种新型的适用于阵列的纠双错编码——ＥＶＥＮＯＤＤ码，以及两种考虑阵列支撑硬件容错能力的阵列结构：正交结构和Ｃｒｏｓｓｈａｔｃｈ结构；指明将纠双错编码应用于正交结构或Ｃｒｏｓｓｈａｔｃｈ结构的盘阵列中，可大大提高阵列可靠性。
After comparing the reliability of Redundant Arrays of Inexpensive Disks(RAID)for tolerating one disk failure with that for tolerating double disk failures,it is found out that the RAID Mean Time To Failure(MTTF)for tolerating double disk failures is thousands of times for tolerating single disk failure if the reliability is considered only for the disk drivers in an RAID. A new kind of code,EVENODD,for toleratingdouble disk failures is described.And two highly reliable architectures are described too,which are orthogonal RAID and Crosshatch RAID.The reliability of RAID will be highly increased if the codes for tolerating double disk failures are used in the orthogonal RAID or the Crosshatch RAID.Ascheme in which the nomal disk driver(not a two portdisk driver)are used to construct a Crosshatch RAID is suggested in the paper.
出处
《计算机工程与科学》
CSCD
1996年第3期43-47,共5页
Computer Engineering & Science
