Efficiently Updating Cost Repository Values for Query Optimization on Web Data Sources in a Mediator/Wrapper Environment

Optimizing accesses to sources in a mediator/wrapper environment is a critical need. Due to a variety of reasons, relational-based optimization techniques are of no use when having to handle HTTP-based web sources, so new approaches which take into account client/server communication costs must be devised. This paper describes a cost model that stores values from a complete set of web source-focused parameters obtained by the web wrappers, by using a novel updating technique that handles the values measured by the wrappers in previous query executions, and generates a new model instance in each new iteration with an efficient processing cost. This instance allows rapid value updates caused by changes of the server quality or bandwidth, so typical in this context. The results of these techniques are demonstrated both theoretically and by means of an implementation showing how performance improves in real-world web sources when compared to classical approaches.