Much of the database research of recent years was aimed at providing an efficient processing method of spatial queries in the object-oriented database management system(DBMS) environment. The query optimizer is an important part of the spatial DBMS for improving performance, however its field is relatively immature.
This thesis presents the technique of the query optimizer which performs integrated query optimization for both spatial and non-spatial data. This query optimizer provides an equal opportunity for both of these two data types to participate in query processing and optimization. It has to be extensible, because each application domain has different optimization requirements. This extensibility provides easy customization of the optimizer.
It is designed based on the Volcano Query Optimizer Generator. The Volcano Query Optimizer Generator is a data model independent tool that is used to develop a query optimizer for a DBMS. The query optimization process is designed in such a way that algebraic manipulations and transformations are performed by considering the effect of the cost analysis on them. In this thesis, a cost model for both spatial data and non-spatial data is developed. Especially, a new cost model for the spatial join based on the Tree Matching join algorithm is proposed. The integrated query optimization technique for both spatial and non-spatial data improves the performance of queries, because the optimizer can choose the optimal execution plan for a query which contains spatial operations and non-spatial operations.