A Methodology for Query Processing over Distributed XML Databases

The constant increase in the volume of data stored as native XML documents makes fragmentation techniques an important alternative to the performance issues in query processing over these data. Fragmented databases are feasible only if there is a transparent way to query the distributed database, without the need of knowing the fragmentation details and where each fragment is located. This paper presents our methodology for XQuery query processing over distributed XML databases, which consists on the steps of query decomposition, including the query’s TLC algebra representation; data localization; global optimization; global query execution and final result assembly. This methodology can be used in an XML database that allows fragmentation and also in a system that publishes an integrated view of semi-autonomous and homogeneous XML databases. We propose an architecture based on a Mediator with Adaptors (wrappers) attached to remote databases. The Mediator publishes a global XML view of the distributed data, which can be queried by users in a transparent way. A Mediator and two Adapters prototypes have been implemented and experiments were executed, where we could analyze the performance improvements and impacts of different queries over distributed XML databases.