【摘要】 目前,已有LRFU(Least Recently Frequently Used)方法结合了访问时间和访问次数来优化缓存,但却无法适用于操作系统、存储系统、web应用等复杂场景。为了解决LRFU算法中无法动态调整λ以及现有自适应调整算法无法兼顾多种访问模式的问题,本文提出了一种基于CAR(Clock with Adaptive Replacement)动态调整策略的改进LRFU算法——CLRFU,并将该算法与局部性定量分析模型相结合,能够在不同访问模式下动态调整λ。实验结果表明,CLRFU算法在线性、概率和强局部访问模式下都具有较好的适应性,提高了缓存整体命中率。更多还原

【Abstract】 At present,existing LRFU( Least Recently Frequently Used) method is a combination of access time and the number of access to optimize cache,but it will not apply to operating systems,storage systems,web applications and other complex scenes. In order to solve the LRFU algorithm in dynamic adjustment λ and the existing adaptive algorithm can’t combine multiple access mode,this paper proposes a improved LRFU algorithm- CLRFU which is based on the CAR( Clock with the Adaptive Replacement),with local quantitative analysis model,the combination of dynamic adjustment λ to different access mode. The experimental results show that the CLRFU algorithm has good adaptability in linear,probability and the strong local access mode and improve the cache hit ratio as a whole.更多还原