Querying XML Data by Parallel Holistic Twig Joins Processing

XML has become the de facto standard for data representation and exchange over the Internet. Nevertheless, along with the increasing size of XML documents and complexities to evaluate XML queries, existing query processing performance in a single-centralized environment will deteriorate. Parallelism is, thus, a viable solution. One possible approach of exploiting parallelism is to adopt a PC cluster system, in which tens of commodity PCs are interconnected with a high-speed network, due to its recent popularization and commercialization. The core operation of XML query processing is to find query patterns in XML data. One of the techniques to find query patterns is the holistic twig joins algorithms, which are important family algorithms to enable us to process queries consisting branches holistically. To the best of our knowledge, parallel query processing based on holistic twig joins has not been studied very intensively. At this stage of the project we focus on finding methods of partitioning XML data and distributing them to cluster PCs. The first method, Grid Metadata for XML (GMX), aims at partitioning XML data obtained from heterogeneous XML documents for inter query parallelism. The second method, streams-based partitioning, aims at partitioning XML data on the fly for intra query parallelism.