摘要
对于数据重构,纠删码提供了一个特定的编码方法,用于保护那些在磁盘阵列中的多重失效。在RAID的应用中,用纠删码为条块数据丢失建模,以便优化重构算法来重构整个条块。换句话说,它们只应用于高度相关的扇区故障,也就是在丢失磁盘上连续的扇区。定位了两个更一般的问题:①由分散或不相关的擦除导致丢失的数据的恢复;②由单个磁盘(存在许多故障时)导致的部分但连续的丢失数据的恢复。对两个问题所建议的方法是完全一般化的,并且能够应用于任何纠删码,但是此方法最适合基于异或的编码。对于分散的擦除,典型的是为每一个丢失扇区的数据规定了两种结果:要么这个丢失的数据被宣布为不可恢复,要么宣布为可恢复。并且,为只依赖于可读扇区的重构提供一个规则。简而言之,这个方法既完整又具有建设性。
Erasure code provides a specific coding method for data reconstruction which is for the protection of those multiple failure in the disk array. In RAID applications,with Erasure code strips for data loss modeling to optimize reconstruction algorithm to the entire strip reconstruction. In other words, they are only applied to a high degree of faultrelated sectors which is the sequential sector on the lost disk. We addressed two more general issues： data recovery from scattered or not related erase sectors which lead to the loss;data recovery from single disk failure（existing many faults） which lead to the partial loss. The two issues of the proposed method are completely generalized and can be used in any Erasure code,but best suited to XOR-based Erasure code. For scattered erasure code, we typically provide for two resuits for every loss sector data., either the lost data is declared unrecoverable or it is declared recoverable and a formula is provided for the reconstruction that depends only on readable sectors. In short, the methodology is both complete and constructive.
出处
《计算机科学》
CSCD
北大核心
2009年第7期262-266,294,共6页
Computer Science
基金
国家自然科学基金项目(60603074)资助
