云环境下一种改进的自适应缓存替换算法  

An Improved Adaptive Cache Replacement Algorithm for Cloud Environment

作　　者：马亚铭 王兆俊 石郡儒 Ma Yaming;Wang Zhaojun;Shi Junru(Institute of China Academy of Space Technology,Beijing 100095,China)

机构地区：[1]中国空间技术研究院第503研究所,北京100095

出　　处：《信息与电脑》2020年第3期27-31,共5页Information & Computer

摘　　要：传统的缓存替换策略大多是针对单机中分页内存管理系统提出的,随着云计算的出现和发展,各种应用纷纷部署到云环境中,共用相同的物理缓存空间。在云环境下缓存系统面临更复杂的负载和多变的访问模式,传统的替换策略并不能很好地发挥作用。因此,各种自适应的缓存替换算法陆续被提出。ACME(Adaptive Caching Using Multiple Experts)是一种可以自动根据当前缓存负载变化,选择最佳的替换算法的自适应算法,较之传统的缓存替换算法,可以在复杂访问模式下,进一步提升缓存命中率。但其自身还存在一些缺陷,在应用到生产环境中时还存在一些问题,笔者将针对这些问题对ACME算法进行改进。实验结果表明,改进的ACME算法可以获得更高的缓存命中率。Most of the traditional cache replacement strategies are proposed for the paging memory management system in a single machine.With the emergence and development of cloud computing,various applications have been deployed to cloud environments,which share the same physical cache space.In the cloud environment,the cache system is facing with more complicated loads and changeable access modes,and the traditional replacement strategy can not work well.Therefore,various adaptive cache replacement algorithms have been proposed.ACME(Adaptive Caching Using Multiple Experts)is an adaptive algorithm that can automatically select the best replacement algorithm according to the current cache load change.Compared with the traditional cache replacement algorithm,it can further improve the cache hit rate in complex access modes.However,there are still some defects in ACME,and some problems in the application to production environment.In this paper,the ACME algorithm is improved to solve these problems.Experimental results show that the improved ACME algorithm can achieve higher cache hit rate.

关 键 词：云环境 缓存替换算法 自适应 复杂负载 

分 类 号：TP319[自动化与计算机技术—计算机软件与理论]