Adaptive Cache Replacement:A Novel Approach

Abstract

Cache replacement policies are developed to help insure optimal use of limited resources. Varieties of such algorithms exist with relatively few that dynamically adapt to traffic patterns. Algorithms that are tunable typically utilize off-line training mechanisms or trial-and-error to determine optimal characteristics. Utilizing multiple algorithms to establish an efficient replacement policy that dynamically adapts to changes in traffic load and access patterns is a novel option that is introduced in this article. A simulation of this approach utilizing two existing, simple, and effective policies; namely, LRU and LFU was studied to assess the potential of the adaptive policy. This policy is compared and contrasted to other cache replacement policies utilizing public traffic samples mentioned in the literature as well as a synthetic model created from existing samples. Simulation results suggest that the adaptive cache replacement policy is beneficial, primarily in smaller cache sizes.