Timo Boehme of the database group at the University of Leipzig has announced XMach-1, a scaleable, multi-user benchmark for XML data management systems.

The benchmark is designed to measure performance of XML data management systems in a Web environment and considers different types of XML documents, including document-centric, data-centric, and schema-less documents. It is designed to be used with both native XML databases and XML-enabled databases.

The benchmark specifies the structure of the database and describes how to generate its contents. It consists of seven retrieval queries as well as insert, delete, and update queries, mixed to simulate a typical Web application. The retrieval queries simulate both data-centric and document-centric operations, and range from simply retrieving a document to full-text searches to queries requiring operations such as joins, grouping, counting, and sorting.

Because most databases that support XML today do not support XML query languages, the benchmark architecture includes an application over the database. This can be a thin layer over a query processor (for databases that support XML queries) or a much thicker layer that hard-codes the logic needed to perform the benchmark queries (for databases that do not support XML queries).

The authors are currently implementing the benchmark over various databases, as well as building a tool to populate a benchmark database.