The world wide wait: where does the time go?

The continuing explosive growth of the web has not been matched by an adequate enhancement of the infrastructure on which it depends. Both consumers and producers are often left frustrated at the disappointingly slow speed of service. Yet there has been little systematic attempt to analyse the sources of delay experienced by users. We present a structured timing model which covers the entire period from a user clicking on a URL to the resultant display becoming usable. The model is not predicated on any particular server, browser or networking infrastructure and provides a basis for discovering the constituent parts of the total delay as experienced by the user in any given situation. Only when the delay is broken down is it possible to make productive decisions about tuning systems and replacing hardware or software components. We illustrate the use of the model with reference to understanding and reducing delays encountered in Finesse, a widely distributed and highly interactive teaching and learning environment.