Handling sharable queries in both streaming and stored XML documents

Abstract

XML Publish/Subscribe systems are system-active and user-passive systems that asynchronously serve users' queries. These queries are usually stored and later executed when matching XML documents become available. With large amounts of subscriptions and in the presence of sharable queries, efficient execution of these queries becomes a substantial requirement. In this work, we describe a novel approach that uses new physical algebra which is capable of handling a group of sharable queries that are submitted against streaming and/or stored XML documents. Our grouping approach is designed to share computations and eliminate redundant executions by producing a single algebraic query execution plan (QEP) for all resident queries and sharing execution in composed physical operators. To assess the effectiveness of our approach, we designed and implemented the physical algebra operators in a system which we then used to conduct experiments that helped us to compare between processing a single query and that of group of queries. Details of our proposed approach, the design and implementation of the related system and the results of our experimental study are presented in this article, together with a discussion of planned directions of future work.