XML View Indexing Using an RDBMS based XML Storage System

Caching query results and reusing them in processing of subsequent queries is an important query optimization technique. Materialized view and view indexing are the representative examples of such a technique. The two schemes had received much attention for relational databases, and have been investigated for XML data since XML emerged as the standard for data exchange on the Web. In XML view indexing, XML view xv which is the result of an XML query is represented as an XML view index(XVI), a structure containing the identifiers of xv's underlying XML elements as well as the information on xv. Since XVI for xv stores just the identifiers of the XML elements not the elements themselves, when xv is requested, its XVI should be materialized against xv's underlying XML documents. In this paper, we address the problem of integrating an XML view index management system with an RDBMS based XML storage system. The proposed system was implemented in Java on Windows 2000 Server with each of two different commercial RDBMSs, and used in evaluating performance improvement through XML view indexing as well as its overheads. The experimental results revealed that XML view indexing was very effective with an RDBMS based XML storage system while its overhead was negligible.