一种应用于分布式缓存系统中的缓存部署算法

缓存技术能有效的节省网络带宽,减少用户的访问延迟.在分布式缓存系统中,一个值得研究的问题是如何根据用户的请求动态的进行缓存部署,使得系统的收益最大.描述了缓存部署问题并建立了优化模型,在此基础上提出一种新的协作缓存部署算法,该算法利用对象的热度、网络距离,以及系统中各节点接收的请求和系统缓存分布信息,依次对请求路径上的节点进行缓存部署决策,同时该算法将计算分布到请求路径的各个节点上进行.仿真结果表明,该算法具有比LRU和Graph算法更高的缓存命中率和更低的访问延迟.

Cache Deployment Algorithm in Distributed Caching System

Cache techniques can be used to save network bandwidth,reduce user access latency.One important problem in distributed caching system is how to deploy cache dynamically,according to the user′s request,to maximize the system′s total benefit.This paper first describes the cache depoly problem,and builds the optimization model for this problem,then proposes a new cooperated cache deployment algorithm,the algorithm utilizes the object popularity,network distance,the request information that every node receive and system cache distribution information,every node that on the request path make a decision on cache deployment in sequence,meanwhile,this algorithm distributes the calculation to the nodes on the request path.As the simulation result shows,this new algorithm has higher hit ratios and lower access latency than LRU and Graph Algorithm.