Least popularity per byte replacement algorithm for a proxy cache

Abstract

With the recent explosion in usage of the world wide web, the problem of caching web objects has gained considerable importance. Web caches cannot only reduce server load, network traffic and downloading latency by replicating popular web objects on proxy caches. The performance of these web caches is highly affected by the replacement algorithm. Today, many replacement algorithms have been proposed for web caching and these algorithms use the other on-line fashion parameters like size, temporal locality and latency to define the object popularity rather than the object popularity value directly from the cache, especially in the Size Adjust LRU which uses size and temporal locality. But, recent studies suggest that the correlation between the on-line fashion parameters, especially temporal locality and the object popularity in the proxy cache is weakening due to the efficient client caches. In this paper, we suggest a new algorithm, called Least Popularity Per Byte Replacement(LPPB-R). This LPPB-R algorithm is the extension of the Size Adjust LRU. We use the popularity value as the long-term measurements of request frequency to complement the weak point in the temporal locality and vary the popularity value by changing the impact factor easily to adjust the performance to needs of the proxy cache. In addition, we apply the multi queue by managing the objects and the meta information and suggest a technique for managing objects to avoid the cache pollution phenomenon. And we examine the performance of this and other replacement algorithm via trace driven simulation.