Pathfinder: Relational XQuery Over Multi-Gigabyte XML Inputs in Interactive Time; Information systems rept

摘要

Using a relational DBMS as back-end engine for an XQuery processing system leverages relational query optimization and scalable query processing strategies provided by mature DBMS engines in the XML domain. Though a lot of theoretical work has been done in this area and various solutions have been proposed, no complete systems have been made available so far to give the practical evidence that this is a viable approach. In this paper, we describe the relational XQuery processor Pathfinder that has been built on top of the extensible RDBMS MonetDB. Performance results indicate that the system is capable of evaluating XQuery queries efficiently, even if the input XML documents become huge. We additionally present further contributions such as loop-lifted staircase join, techniques to derive order properties and to reduce sorting effort in the generated relational algebra plans, as well as methods for optimizing XQuery joins, which, taken together, enabled us to reach our performance and scalability goals.