用双目标加权遗传算法解决网络磁盘阵列系统下校验散布布局优化问题的研究  

Solving Parity Declustering Optimal Problems in Network Based RAID by Using Two-Weighted-Objective Genetic Algorithms

作　　者：杨敏[1] 王刚[1] 刘璟[1] 陈北莲[1] 

机构地区：[1]南开大学计算机科学与技术系,天津300071

出　　处：《计算机科学》2005年第5期73-75,100,共4页Computer Science

基　　金：国家自然科学基金(项目编号60273031);高校博士学科点专项科研基金(项目编号20020055021)

摘　　要：廉价磁盘冗余阵列(RAID)作为一种提高存储系统可靠性和性能的技术,已经得到了广泛的应用,有关磁盘阵列结构和数据布局的研究也一直很活跃,但有关网络磁盘阵列下的数据布局的研究还不太多。本文首先概述了校验散布布局的技术和遗传算法的相关知识,提出了利用双目标加权遗传算法的思想解决网络磁盘阵列系统校验散布布局优化的问题。然后以“重构负载均匀分布”和“校验均匀分布”为双目标,使用改变的NSGA来解决网络磁盘阵列系统下校验散布布局的优化问题。最后给出了实验结果。RAID,a technology to improve the reliability and performance of storage system,has been got a widely us- ing. Although the researchers about the structure of RAID and data layout are very popular, the researchers about the data layout of network based RAID are little. In this paper, first, we mainly introduce some knowledge about parity declustering and genetic algorithms, present a new thought to solving parity declustering optimal problems in the network based RAID by two-weighted-objective genetic algorithms. Then we take 'Balance reconstruction work- load' and 'Balanced parity overhead'as two objectives, used the changed NSGA to solving parity declustering optimal problems in the network based RAID system. At last we describe the experimental results.

关 键 词：磁盘阵列系统 遗传算法 布局优化问题 双目标 散布 加权 网络磁盘阵列 磁盘冗余阵列 数据布局 均匀分布 系统可靠性 阵列结构 系统校验 技术 

分 类 号：TP333.35[自动化与计算机技术—计算机系统结构] O242.23[自动化与计算机技术—计算机科学与技术]