集群VOD系统中磁盘Cache替换算法研究

Research on Disk Cache Replacement Algorithms for Clustered Video-on-demand System

根据VOD的特点开发了两种基于访问频率的替换算法:LFRU(least frequency and recently used)和PLFU(period least frequency used)算法，它们都试图将访问频率大的视频数据保留在Cache中。LFRU算法结合了数据的访问频率和访问时间信息，对访问模式的变化具有一定的适应性。PLFU算法用周期法和预测法解决了LFU算法中的Cache“污染”问题。

In this paper, two new Cache replacement algorithms based on access frequency: LFRU (least frequency and recently used) and PLFU (period least frequency used) are developed. LFRU is a trade-off between LRU and LFU, which has advantages of the both algorithms. PLFU is an improved LFU algorithm, which solves the Cache pollution problem of LFU algorithm.