University of Waterloo
Database Generator
Schemas and DTDs
Other XML Benchmarks
School of Computer Science
XBench -
A Family of Benchmarks for XML DBMSs

The workload of an XML benchmark should include queries, updates and bulk loads. In this first version of XBench, only queries are considered. These queries challenge a DBMS with XML-specific features as well as conventional DBMS functionalities. The workload is classifed along two dimensions: by functionality and by document class. All queries are expressed in natural language English, followed by XQuery expression. The workload subsumes all of XQuery's functionality, as captured by the XML Query Use Cases.

The following table lists all queries with respect to the features that they are supposed to test.

Queries Functionality
...Q1 Top level exact match
...Q2 Deep level exact match
...Q3 Function application
...Q4 Relative ordered access
...Q5 Absolute ordered access
...Q6 Existential quantifier
...Q7 Universal quantifier
...Q8 Regular path expressions (unknown element name)
...Q9 Regular path expressions (unknown subpaths)
...Q10 Sorting by string types
...Q11 Sorting by non string types
...Q12 Document structure preserving
...Q13 Document structure transforming
...Q14 Missing elements
...Q15 Empty (null) values
...Q16 Retrieve individual docs
...Q17 Uni-gram search
...Q18 N-gram search
...Q19 References and joins
...Q20 Datatype Cast

For each document class, there is a set of domain specific queries, which can be found at the [Downloads] section.

University of Waterloo
University of Waterloo
School of Computer Science
School of Computer Science
Database Research Group
Database Research Group

Benjamin Bin Yao
Last modified: Mon Dec 9 09:11:20 EST 2002