结合流行度选择的集群网络高速缓存优化仿真  

Cache Optimization Simulation of Cluster Network Combined with Popularity Selection

作　　者：赵永乐 吴坤芳 ZHAO Yong-le;WU Kun-fang(Luohe Institute ofIndustry,Henan University of Technology,Luohe Henan 462000,China)

机构地区：[1]河南工业大学漯河工学院,河南漯河462000

出　　处：《计算机仿真》2020年第9期262-265,共4页Computer Simulation

摘　　要：近年来,集群网络逐渐由以主机为核心的模式向内容模式演变,网络内容缓存受到了越来越多的关注。针对集群网络海量内容的缓存性能问题,提出结合流行度选择替换的高速缓存优化方法。为了准确评估各个节点的缓存压力,根据节点的缓存占用率计算得到所有节点中的负载情况,同时利用网络中各个内容的流行度对其进行优先权划分,结合内容大小以及在节点处的缓存基本优先级来评估具体内容在节点处的缓存优先级,进而根据节点处的缓存优先级别,计算出请求内容被替换的概率,将流行度低的缓存内容进行替换,从而保证缓存提供有效的内容流行度信息。仿真结果表明,上述存缓优化方法具有较高的缓存命中率,以及高效的缓存替换率,大大提升了集群网络的高速缓存性能。In recent years,the cluster network has gradually evolved from the host centered model to the content model,and the network content caching has attracted more and more attention.Aiming at the cache performance of massive content in cluster network,a cache optimization method combining popularity selection and replacement is proposed.In order to accurately evaluate the cache pressure of each node,the load of all nodes is calculated according to the cache occupancy rate of each node.At the same time,the popularity of each content in the network is used to divide its priority,and the size of the content and the basic priority of the cache at the nodes are combined to evaluate the priority of the specific content at the nodes.Then,according to the priority of the cache at the node,the probabili⁃ty of the request content being replaced is calculated,and the cache content with low popularity is replaced,so as to ensure that the cache provides effective content popularity information.The simulation results show that the cache op⁃timization method has high cache hit rate and efficient cache replacement rate,which greatly improves the cache per⁃formance of the cluster network.

关 键 词：集群网络 缓存压力 内容流行度 替换率 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]