AdpReplica:自适应副本管理机制  

AdpReplica:Adaptive Replica Management Mechanism

作　　者：魏青松[1] 卢显良[1] 侯孟书[2] 

机构地区：[1]电子科技大学计算机学院,成都610054 [2]同济大学软件学院,上海200331

出　　处：《计算机科学》2004年第12期34-36,51,共4页Computer Science

基　　金：国家95重点攻关项目支持

摘　　要：本文提出一种自适应副本管理机制—AdpReplica。AdpReplica将期望可用性和一致性维护开销看成副本数的函数,并建立自适应模型。该模型动态管理副本,使文件副本数维持在一个合理的水平,既满足用户的期望可用性,提高访问效率和平衡负载,又减少带宽消耗,保证系统稳定,为用户提供满意的存储QoS。In this paper, we present an Adaptive Replica Management Mechanism called AdpReplica. The AdpReplica constructs adaptive model regarding user anticipant availability and consistency maintenance overload as a function of a file's replica number. The AdpReplica model manages replica dynamically and maintains a rational replica number, not only satisfying the user anticipant availability, improving access efficiency and balancing overload, but also reducing bandwidth requirement, keeping the system stable and proving satisfied storage QOS.

关 键 词：访问效率 自适应 一致性维护 可用性 文件 开销 QOS 用户 管理机制 动态管理 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]