摘要
分布式存储系统常常使用纠删码冗余技术提高系统可靠性.为保证一定的冗余度,系统必须具有自修复能力再生失效数据.再生码是纠删码的一种改进形式,最大特点在于无需下载整个数据文件就能恢复单个节点数据,从而有效减少数据修复时的网络带宽.相关文献证明再生码数据修复时存在极值点—最小带宽再生点(MBR),由此提出最小带宽再生码MBRC.文中从数据分布、失效数据修复和数据重构三个方面描述了实现的原理,并通过构建数据矩阵和修复矩阵实现MBRC再生码.利用实例详细给出了再生码的实现过程,并理论证明正确性,最后仿真实验验证了MBRC的有效性.
Distributed storage systems often use redundancy technology to improve system′s reliability,such as erasure codes.In order to ensure a certain redundancy degree,the system must have self-healing ability to regenerate failure data.Regeneration code is an improvement form of erasure codes,which has the feature that no needs to download the entire file to recover a single node data.Therefore,it can effectively reducing the network bandwidth when data recovery.Some literatures prove that there are extreme points when data repair for erasure code,which is the minimum bandwidth regeneration points(MBR).We raised the minimum bandwidth regeneration code(MBRC).In this paper the principle of MBRC is given in detail and the implementation process is described by examples.the final simulation experiments verify the effectiveness of the MBRC.
出处
《小型微型计算机系统》
CSCD
北大核心
2012年第8期1710-1714,共5页
Journal of Chinese Computer Systems
基金
国家自然科学基金项目(60573145)资助
博士点基金项目(200805610019)资助
广东省自然科学基金项目(10151063301000000)资助
广东优秀青年创新人才培育项目(LYM09098)资助
关键词
分布式
存储系统
再生码
数据修复
distributed system
storage system
regeneration code
data repairing