Massively multi-query join processing in publish/subscribe systems

Mingsheng Hong, Alan J. Demers, Johannes E. Gehrke, Christoph Koch, Mirek Riedewald, Walker M. White
2007 Proceedings of the 2007 ACM SIGMOD international conference on Management of data - SIGMOD '07  
There has been much recent interest in XML publish/subscribe systems. Some systems scale to thousands of concurrent queries, but support a limited query language (usually a fragment of XPath 1.0). Other systems support more expressive languages, but do not scale well with the number of concurrent queries. In this paper, we propose a set of novel query processing techniques, referred to as Massively Multi-Query Join Processing techniques, for processing a large number of XML stream queries
more » ... ing value joins over multiple XML streams and documents. These techniques enable the sharing of representations of inputs to multiple joins, and the sharing of join computation. Our techniques are also applicable to relational event processing systems and publish/subscribe systems that support join queries. We present experimental results to demonstrate the effectiveness of our techniques. We are able to process thousands of XML messages with hundreds of thousands of join queries on real RSS feed streams. Our techniques gain more than two orders of magnitude speedup compared to the naive approach of evaluating such join queries.
doi:10.1145/1247480.1247564 dblp:conf/sigmod/HongDGKRW07 fatcat:tec4ctah4bck7dztqepgwkpycy