分布式存储中的再生码综述

Survey for regenerating codes for distributed storage

分布式存储系统中通过引入冗余提高系统的可靠性,纠删码作为重要的冗余策略在分布式存储中得到越来越多的重视。分布式存储系统中,当某个存储节点失效后,需要引入新的节点来修复失效节点的数据。传统纠删码冗余策略在修复失效节点时需要传输的数据量较大。近年来出现的再生码对传统纠删码进行改进以减少修复失效节点的带宽消耗。再生码引入网络编码的思想,在修复失效节点时,参与修复过程的节点首先将本节点内的数据作线性组合后再上传,最终修复带宽消耗最小。介绍了再生码的基本概念,然后介绍单节点修复再生码和合作修复再生码的编码策略,最后总结再生码的发展和研究方向。

Distributed storage system introduces redundancy to increase reliability,and erasure code is used widely in distributed storage system.When a node fails in the distributed storage system,a new node is introduced to repair the failed one.When using erasure code,repairing the failed node often requires a large amount of data transfer.Regenerating Code was proposed to reduce the cost of repair bandwidth.Regenerating Code reduces the amount of data uploaded by each helping node by transferring a linear combination of the data it stores to the failed node in the repair process.In this paper we first review the basic idea of regenerating code.Then we summarize the code constructions of both the one-by-one and cooperative regenerating code.At last we talk about the future research direction.