In this paper we present a technique for reducing response delay for web systems, which is based on a proactive cache scheme. It combines predictive pre-fetching and streaming to overlap the read-time with loading time. This graph-based model analyzes the hyperlink structure to form the prediction. It also utilizes data streaming to further minimize the pre-load, without compromising the responsiveness. The analysis demonstrates that such new hyper-graph based pre-fetching can reduce the lag-time of a cache system by a factor ranging from 2-10. In this paper, we focus on pre-fetching and provide the technique, the optimization algorithms and the simulation results.
Javed I. Khan