Abstract

Abstract. Path expression processing optimization is a central and difficult issue in current query languages. Efficient evaluation of path expressions in a distributed context involves choosing among several query-processing strategies. The great variety of strategies comes from the rich semantics involved in object-based data models and the complexity added by the distribution. This work presents a new cost model for object query processors and addresses relevant issues, which are ignored or relaxed in other works in the literature, such as the selectivity of the path expression, the sharing degree of the referenced objects, the partial participation of the referenced collection in the relationship, object clustering in the disk and the distribution of the database objects among the nodes of a network. These issues allowed us to reduce the error rate between our simulation results and the experimental results obtained in previous works, thus presenting more realistic estimates for the query optimizer. Our cost model has been validated against experimental results obtained with an object DBMS prototype running in a distributed architecture, using the OO7 benchmark application.

Citations