Adaptive Web caching using logistic regression

A fundamental understanding of Web access patterns is necessary before we can tackle cache performance issues. Due to the extremely dynamic nature of the Web, any techniques we employ must be able to adapt to changing conditions on the Web. We introduce a near-optimal cache algorithm, based on complete a priori knowledge of future Web access. In reality, such knowledge is unavailable. To enable us to predict future accesses, we have adopted a logistic regression model. Using this model, a Web caching agent can acquire knowledge about the objects it encounters and deduce an effective strategy dynamically. Preliminary results (based on object hit and byte hit ratios) on trace-driven simulations of five different servers are very promising.