摘要
网格中资源能力的不均衡和异构特性会给系统造成瓶颈,引起客户端延迟,为了提高网格的性能,在网格中引入了cache技术,有效地缓解了这一问题。在cache中没有检索到需要的文件时,网格的SRM如何从网格中以一个最快、最有效的方式来获取一个该对象的副本,并将其置于本地的cache中,是一个迫切需要研究的问题。该文在引入经济学模型的基础上,运用遗传算法,给出了该问题的一个有效解决策略,经过初步分析,该算法可以达到满意的结果。
Disproportions between the capabilities of the resources in the grid and its property of heterogeneity cause bottlenecks in the system and high latency seen by the users.To improve the performance of grid,cache technique has been introduced into grid and the grid cache relaxes this problem effectively.In applications that access large files over wide area network environment,caching mechanism can significantly improve the file access performance under appropriate workloads.When a file is unavailable in acan grid SRM＇s cache,it should determine the fastest and the most cost effective means of fetching a copy into its cache.But how to solve the problem for SRM has not been studied up to the present.This paper analyses the grid cache technology,and gives a scheme which solves the problem based on Genetic Algorithm.Analysis shows that the policy is quite effective.
出处
《计算机工程与应用》
CSCD
北大核心
2005年第22期124-126,163,共4页
Computer Engineering and Applications
基金
国家自然科学基金(编号:60373063)
山东省自然科学基金(编号:Y2002G03)资助
作者简介
宋风龙（1980-），男，硕士研究生，主要研究方向为网格计算、服务网格、并行计算等。E-mail：longdragen@sohu．com刘方爱，博士，教授，博导，主要研究方向为分布式处理、并行算法、光互联网络路由算法、网格计算、网络环境下的应用开发技术等。
