A Cache Replacement Algorithm Based On Byte Hit Ratio

With the rapid development of the Internet, people around the world cancommunicate with each other easily. The huge data it produces is a big challenge tothe network bandwidth. But improvement of network bandwidth is a huge project, andcan not be an effective solution in a short time. Web caching technology is an optionalway to improve QoS (Quality of Service). It can reduce the load on the network,shorten network latency and improve clients’ waiting time.First of all, we introduce the basic principles and research of web cachingtechnology, pointing out the two important aspects of a web cache server—architecture of the system and caching strategy. Then we give a detailed description ofthe existing caching strategies and the performance indicators of them.Secondly, after analyzing at process of cache replacement, we conclude that it’s atypical0/1knapsack problem. Pointing out that the size of cache data, the count it hasbeen requested, the interval it stays in the cache affect it’s value, we present a cachereplacement algorithm based on byte hit ratio. The algorithm estimates the value ofcache data making full use of known information. Also we know that a networkresource is often referenced by different network addresses. Restoring same cachedata, is not only a waste of cache memory, but also will affect the estimation of thevalue of the data. In order to solve this problem, we use the MD5digest algorithm tocompare cache data with another to find whether they are the same. Any cache data isstored once, but it might correspond with more than one network address.Finally, we design a simple model of web cache and use trace-driven simulationto test its performance. By comparing the performance of deferent cache replacementalgorithms, we say that the algorithm presented achieves improvement in byte hitratio.