An Adaptive Buffer Cache Management Scheme

Abstract

Previous cache replacement algorithms utilize the access history information to make replacement decisions. However, they fail to deliver utmost performance since the history information exploited is incomplete. Motivated by the limitations of existing algorithms, this paper proposes a novel replacement scheme, called the Pattern-assisted Adaptive Recency Caching (PARC). PARC simultaneously utilizes the history information of recency, frequency, and access patterns to estimate the locality strength and to select the victim block. Specifically, PARC exploits the reference regularities exhibited in past behaviors, including looping or sequential references, to actively and rapidly adapt the recency and frequency information of blocks so as to exactly distill blocks with long-term utility from those with only short-term utility. Through comprehensive simulations on a variety of traces of different access patterns, we show that PARC is robust since, except for random workloads where the performance of each cache replacement algorithm is similar, PARC always outperforms the least recently used (LRU) scheme and other existing cache replacement algorithms.