一种基于有限记忆多LRU的Web缓存替换算法

Limited History Based Multi-LRU Web Cache Replacement Algorithm

Web缓存的核心是缓存内容的替换算法.在动态不确定的网络环境下,本文提出一种基于有限记忆的多LRU(LH-MLRU)Web缓存替换算法,它是一种低开销、高性能和适应性的算法.LH-MLRU综合考虑各项因素对Web对象使用多个LRU队列进行分类管理,引入Web对象最近被访问的历史作为缓存内容替换的一个关键因素,来预测对象可能再次被访问的概率.通过周期性的训练参数可以适应动态不确定的网络环境.轨迹驱动的仿真实验表明LH-MLRU在各项性能指标上均优于其他算法,可以显著的提高Web缓存的性能.

A core of Web caching is the replacement policy of Web cache. In a dynamic and uncertainty Web environment,this paper introduces a based on Limited History Multi-LRU （LH-MLRU） algorithm for Web cache, which is a low overhead and high performance and adaptive algorithm. LH-MLRU generally considers various factors and uses Multi-LRU to manage Web objects. Furthermore, it introduces the limited history of references to Web objects as a critical parameter for the cache content replacement to predict the probability of revisit to we objects. Through training parameters periodically it can be adapted to dynamic and uncertainty Web environment. Trace-driven simulations with actual Web cache logs show that LH-MLRU outperforms other policies for various performance measures, and can significantly improve the performance of Web cache.