企业信息分布式存储的热点处理策略  

Hotspot treatment strategy of distributed storage for enterprise information

作　　者：王征[1] 刘心松[1] 李美安[1] 

机构地区：[1]电子科技大学计算机科学与工程学院,四川成都610054

出　　处：《计算机集成制造系统》2006年第9期1506-1510,1515,共6页Computer Integrated Manufacturing Systems

基　　金：四川省应用基础研究资助项目(04JY029-017-2)~~

摘　　要：为解决企业信息分布式存储中因局部访问过热而导致的一系列问题,提出了新型的分布式存储热点处理策略。该策略将存储对象的热度划分为直接热度、内部热度和间接热度,并且根据该划分提出不同的热点处理子策略,建立了一个企业分布式存储系统的热点处理模型。根据该模型及策略,给出了与之对应的算法,包括信息收集、选择、协商以及扩散4个子算法。分析和应用结果表明,该策略能够减少内部通信量、缩短用户请求延迟、提高系统有效用户。最后,采用该策略的实例验证了它的有效性。In order to deal with the local hotspot caused by extreme access in enterprise information distributed storage, a novel distributed storage hotspot treatment strategy was presented. The strategy classified the temperatures of storage objects as direct temperature, internal temperature and conductive temperature. Hotspot treatment sub-strategies were respectively proposed according to the classification and a hotspot treatment system model was constructed. Based on them, the implementation algorithm including 4 sub-algorithm of information collection, selection, negotiation and extension was provided. Analysis and application results indicated that the strategy could reduce communication flux within storage systems, cut response delay for the requests from users and increase the effective capability for clients. Finally, the advantages of the strategy were showed by a distributed transparent storage system application.

关 键 词：企业信息 分布式存储 热点处理 定向退火 

分 类 号：TP393[自动化与计算机技术—计算机应用技术] TP311[自动化与计算机技术—计算机科学与技术]