面向集群环境的高可用分布并行存储模型研究设计与实现  

Research and Design of Highly Availability Storage Model Aims at Distributed Cluster

作　　者：夏戈明[1] 黄遵国[1] 

机构地区：[1]国防科技大学计算机学院,湖南长沙410073

出　　处：《计算机应用研究》2006年第7期67-69,共3页Application Research of Computers

基　　金：国家"863"计划资助项目(2003AA142080)

摘　　要：针对分布式集群环境,提出了实现高可用设计的新的设计原则,着重关注相继失效问题和冗余问题。对核心的数据分布算法进行了设计和论证,并在此基础上设计了以数据副本和校验计算相结合的并行分布存储模型,在满足并行性要求的前提下,以适度的时空开销实现了多级容错能力的高可用存储。Take account of the circumstance of distributing cluster, this paper proposes a few new principles to get high availability, especially looking at the questions of failures one after another and redundancy. Basing on the designing and demonstrating of data distributing algorithm, the key matter, a parallel distributing storage model was realized using both copying and checkout, which satisfies parallel and multilevel fault tolerance , along with appropriate spending of space-time.

关 键 词：高可用 冗余 多级容错 相继失效 存储模型 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]