Content area
The thesis is concerned with two query processing and optimization problems in relational database systems, namely, the distributed query processing problem in broadcasting local area networks and the multiple query optimization problem. In this thesis, a systematic way is proposed to deal with these two problems. For a given query or a set of queries, a query state is defined. Each query processing operation (selection, join etc.) is treated as a query state transition. So the query optimization problem can be considered as a problem of searching for a minimum cost path from the original query state to the final query state.
In the distributed query processing environment, semijoin has been used to reduce the data communication cost. For Broadcasting Local Area Networks (BLAN's), communication cost and local processing cost are comparable, semijoin may not be as useful as in long-haul networks. The conventional two-phase strategy (semijoin reduction phase and final join phase) for long-haul networks which emphasizes reducing the relation sizes before joins may not be suitable for local area networks. To deal with the problem of inaccuracy of the database statistics information and to reduce the size estimation errors, we investigate the utility of dynamic query processing methods for BLAN's. We show the advantage of dynamic strategy over static strategy for simple queries. To investigate the performances for general queries, we perform simulations and compare static and dynamic query processing methods. The simulation results show that the new algorithm is an improvement to the existing methods for distributed query processing, and dynamic strategy is superior to the static strategy when the attribute value distribution is very biased.
We also study the problem of multiple query optimization. After a survey of the previous work, we propose a multiple query processing system to optimize a set of queries simultaneously. The system consists of four components: Query analyzer, Query reorganizer, Query optimizer and Query processor. Both general searching methods and heuristic methods are discussed. Two kinds of multiple operations, namely, multiple selection and multiple joins, are introduced to combine operations and further reduce the query processing costs at the physical level. (Abstract shortened with permission of author.)