Benchmarking of native XML database systems

In this thesis, XML database systems, standard database benchmark techniques, and XML database system benchmarks such as XBench, XMark, XMach-1, and X007 are reviewed. A new benchmark system called MyBench is provided for benchmarking of native XML database systems. First, graph grammars are utilized to define a group of productions with different tree structures, then the algorithm for generating XML documents is provided. Second, a group of benchmark query patterns with different functionalities is provided. We also depict how to automatically generate benchmark queries in terms of these patterns and the parameters input by the user. Third, we utilise XSLT to implement MyBench XML document generator, and we also utilise relational DBMS to implement benchmark queries. Finally, MyBench is compared with other XML database benchmark techniques. We also summarise the contributions of this thesis and formulate some open problems.