Enhancing QoS In Web Caching Using Differentiated Services

Web caching and Content Distribution Networks (CDNs) occupy strategic positions in the growth of Internet by effectively delivering data to the end users with reduced access latency. Due to diverse nature of applications, Quality of Service (QoS) is considered an important aspect in web caching. To achieve QoS in Caching scenario web requests are classified into different classes with each class holding different priority levels. In this paper, we show how differentiated strategies combined with dynamic memory allocation based on relative hit rate can provide improved performance for high priority classes with little or no performance degradation for low priority classes. We have developed and evaluated a model that classifies web requests into two classes (Premium, Best-Effort) based on object size that uses LRU and LFU-DA as replacement algorithms. The proposed model efficiently converge the hit rate of classes towards its specified desired hit rate with minimum overhead.