Dynamically Adjusted Reduced Cache Digest Algorithm Based On Sizes Of Requested Objects

Web caching which is generally considered one of the most effective ways to improve the network quality of service can reduce Internet flow, improve the response time of terminal user and reduce network delay. Web caching makes wide transmission to a local request, increasing an additional layer of hierarchical structure. According to some strategies, cache saves a copy of the content which client requests. When the same request comes, the cache can provide the copy to meet the user, instead of asking the origin server again. This technology can reduce repeated transmission and optimize the web bandwidth and traffic.Based on analyzing principle and research trends of web caching, we build an improved method of Reduced Caching Digest through dividing cache catalogue based on sizes of objects in this paper. Reduced Caching Digest (RCD) is a new method based on Cache Digest. RCD divides the cache catalogue into two parts, one of the parts saves the objects which are requested recently and the other part saves the objects which are requested frequently. This division can reduce the space of storage and the bandwidth while issuing the cache catalogue and can alleviate the limitation of the Cache Digest. But the size of the dividing parts of RCD cache catalogue is fixed. Along with large numbers of the clients' requests coming, the contents of cache catalogue are updated ceaselessly. Because size is fixed, when clients' requests are increasing, the frequent part may be filled quickly and objects may be replaced. At the same time the hit ratio of cache descend acutely, and the performance of cache is decreased in a certain degree.In order to increase the hit ratio of web caching, we realize dynamic adjust between recent part and frequent part in improved algorithm. The objects which different users request differ largely. Generally, small objects can be replaced by large ones. Objects which have different sizes cause different requested states. The sizes of requested objects often focus on a certain scope. Objects of other sizes are requested less. So this paper considers the sizes of requested objects when dividing the cache catalogue in order to reduce the defects of RCD.Experiment indicates that dividing the cache catalogue according sizes of objects and adjust each part dynamically can improve the cache performance largely and provide better service.