View article

The data volume of XML repositories and the response time of query processing have become critical issues for many applications, especially for those in the Web. An interesting alternative to improve query processing performance consists in reducing the size of XML databases through fragmentation techniques. However, traditional fragmentation definitions do not directly apply to collections of XML documents. This work formalizes the fragmentation definition for collections of XML documents, and shows the performance of query processing over fragmented XML data. Our prototype, PartiX, exploits intra-query parallelism on top of XQuery-enabled sequential DBMS modules. We have analyzed several experimental settings, and our results showed a performance improvement of up to a 72 scale up factor against centralized databases.