PowerDB-XML: Scalable XML Processing with a Database Cluster

Abstract

The flexible data model underlying the W3C XML document format covers a broad range of application scenarios. These scenarios can be categorized into data-centric and document-centric ones. Data-centric processing stands for highly structured XML documents, queries with precise predicates, and workloads such as the ones with online transaction processing. Document-centric processing in turn denotes searching for relevant information in XML documents in the sense of information retrieval (IR for short). With document-centric scenarios, XML documents are typically less rigidly structured and queries have vague predicates. Today, different systems address these needs, namely database systems and information retrieval systems. XML however offers the perspective to cover them with a single integrated framework, and to make the above distinction obsolete at least in terms of the underlying system infrastructure. The aim of the PowerDB-XML engine being developed at ETH Zurich is to build an efficient and scalable platform for combined data-centric and document-centric XML processing.