A class-based least-recently used caching algorithm for World-Wide Web proxies

Abstract

In this paper we study and analyze the influence of caching strategies on the performance of WWW proxies. We propose a new strategy, class-based LRU, that works recency- as well as size-based, with the ultimate aim to obtain a well-balanced mixture between large and small documents in the cache, and hence, good performance for both small and large object requests. To achieve this aim, the cache is partitioned in classes, each one assigned to a specific document size range; within a class, the classical LRU strategy is applied. We show that for class-based LRU good results are obtained for both the hit rate and the byte hit rate, if the size of the classes and the corresponding document size ranges are well chosen. The latter is achieved by the use of a Bayesian decision rule and a characterisation of the requested object-size distribution. In doing so, class-based LRU is an adaptive strategy: a change in request patterns results, via a change in the distributions, in a change in cache partitioning and request classification. Finally, the complexity of class-based LRU is comparable to that of LRU and, therefore, smaller then of its "competitors". © Springer-Verlag Berlin Heidelberg 2003.