针对现有的副本放置中存在的问题,基于p-中值选址的方法,提出了一种动态的多副本放置算法.该算法预先设置副本的数量,根据网络带宽、存储空间和文件请求次数等条件,求解目标函数——全局费用的最小化,从而确定副本的位置,并随着网格环境的变化而调整.采用Optor-Sim模拟器实验,分析了副本数量、存储空间与副本放置的关系.结果表明:与LRU和基于经济模型的方法相比,该算法能减少数据访问时间和网络利用率.
Based on p-median location, a dynamic algorithm for multi-replica placement was proposed to solve the problem of replica placement. The algorithm regulates the numbers of replica, and minimizes the objective function (total cost) by Computing functions of network bandwidth, sizes of storage space, and the number of documents requested, finally decides the locations of replica and adjusts it according to the environment dynamically. Using the Optor- Sire simulator we analyze the effects of replica numbers and storage spaces on replica replacement. The results show that the algorithm can reduce data transmission times and network utilizations compared to the methods of LRU and economy-based model.