Statistics for Cost-Based XML Query Optimization

Cost-based query optimization (CBO) is a very important area for data management in general and for XML data management in particular. For native XML database management systems (XDBMS), CBO techniques are harder than for relational databases, because the underlying tree-based data model is much more complex and the relative order (document order) between XML elements (nodes) matters. In this paper, we present our first ideas on statistics data structures supporting CBO, that is, the mapping of a logical access model with algebraic operators to the physical access model. In our prototype system called XTC (XML Transaction Coordinator), the physical access model is embodied by a toolbox of path processing operators from which the best performing operators have to be selected based on statistic information for the construction of the query execution plan.