Designing a high-performance database engine for the 'Db4XML' native XML database system

Sudhanshu Sipani, Kunal Verma, John A. Miller, Boanerges Aleman-Meza

Abstract: eXtensible Markup Language (XML) is fast becoming the common electronic data interchange language between applications. In this paper, we describe a database engine called ‘Db4XML’, which provides storage for XML documents in native format. Db4XML is a high performance, main memory resident database engine. Db4XML is being used as a testbed for comparing various query evaluation techniques. The use of wild card (*, ?, ‘//’, etc.) in the path expressions of a query allows users to query documents whose structural information is not available. This paper lists different techniques that can be used to evaluate generalized path expressions (GPE) and presents a performance comparison of the same. A preliminary performance study of the effect of using concurrency control techniques on the various query evaluation techniques is also performed. This paper briefly discusses a suitable recovery technique for the database engine.