Network performance implications of variability in data traffic

World Wide Web (WWW) traffic will dominate network traffic for the foreseea ble future. Accurate predictions of network performance can only be achieve d if network models reflect WWW traffic statistics. Through analysis of usa ge logs at a range of caches it is shown that WWW traffic is not a Poisson arrival process, and that it displays significant levels of self-similarity . It is also shown for the first rime that the self-similar variability ext ends to demand for individual pages, and is far more pervasive than previou sly thought. These measurements are used as the basis for a cache-modelling tool-kit. Using this software the impact of the variability on predictive planning is illustrated. The model predicts that optimisations based on pre dictive algorithms (such as least recently used discard) are likely to redu ce performance very quickly. This means that, far from improving the effici ency of the network, conventional approaches to network planning and engine ering will tend to reduce efficiency and increase costs.