Accurate characterization of WWW traffic

World Wide Web (WWW) traffic models are used to generate the representative workloads that can be used for cache design and the study of "what if" scenarios for WWW-servers planning. Such models should incorporate the essential WWW traffic properties, including temporal locality, spatial locality, and popularity. In this paper, we propose a new WWW server traffic model that accurately captures these properties. Our model uses the inter-request distance string, which we found to be more accurate than the previously proposed stack distance approach in capturing traffic properties. Trace-driven simulation is used to demonstrate the goodness of the proposed model.