Introduced as a schema-less, self-describing data representation language, XML (Bray et al. 2000) has rapidly emerged as the standard for information interchange on the World Wide Web. Database researchers have actively participated in developing standards centered on XML, in particular data models and query languages for XML. Many XML query languages have been proposed, and some XML Management Systems (XMS) with those query languages implemented are available for use.
Current XMS can be divided into two categories: XML-enabled databases and native XML databases. XML-enabled databases, typically relational databases, such as DB2 XML Extender from IBM, Informix, Microsoft SQL Server 2000, and Oracle- 8i and -9i (Bourret 2001a; Chang et al. 2000), provide XML interfaces that allow the storage of XML data in their proprietary relational format, and querying and publishing through an XML representation. Such systems are generally designed to store and retrieve data-centric XML documents. On the other hand, native XML databases such as Kweelt (Sahuguet 2001), IPEDO, Tamino, 4SuiteServer, DBDOM, dbXML (Bourret 2001a), and so on either store the entire XML document in text form with limited retrieval capabilities or store a binary model (e.g., Document Object Model) of the document in an existing or custom data store. With so many XMS being developed and proposed, it is necessary to start designing and adopting benchmarks that allow comparative performance analysis of the tools and systems.
The rest of this chapter is organized as follows. The next section gives benchmark specifications. This is followed by a section that describes three existing XMS benchmarks?the XOO7 benchmark, the XMach-1 benchmark, and the XMark benchmark. Finally, the chapter concludes with a brief comparison of the three benchmarks.