Abstract

I will present the lessons learnt and vision derived from designing and implementing a general purpose distributed Event Processing Network (EPN) project at Betfair. Event Processing (EP) is a field that promised a lot but has so far delivered in a limited way, with pragmatism and narrow application domains dominating the commercial landscape. I will argue that for it to truly realise its potential, EP should flow back into the mainstream of computer science, combining the best in distributed heterogeneous systems, workflows, reactive agents and functional programming. We are implementing a system capable of supporting a variety of different use cases as opposed to single use, distributed as opposed to a “box in the corner” solution, expressive and accessible to end users as opposed to controlled by a closed highly-skilled group. The motto of the day is distributed innovation in the enterprise.

BIO : Alex Kozlenkov studied theoretical physics at Moscow State University with emphasis on unified field theories and gravitation, and started his research career in Space Research Institute working in theoretical astrophysics and space instruments with papers on quantum effects in extremely high magnetic fields and gamma-ray bursts, including publications in Soviet Physics Journal, Astronomy and Astrophysics, and Nature. He then worked at P.N. Lebedev Physics Institute focusing on applications of algebraic topology methods to signal analysis problems in aperture synthesis imaging. Further shifting into computer science, he then worked in a startup company Greentech in Sweden, working on computational non-search approaches to NP-complete problems, and developed proprietary algorithms that solved a number of challenge SAT problems including first solution to parity problems. In 2001, he joined the School of Informatics at City University, focusing on specification inconsistency, abduction, macro tree transducers, bioinformatics, and reactive agents. He also pioneered an area of ontology-driven search (eventually resulting in the product GoPubMed (www.gopubmed.com) used widely in bioinformatics). He published papers and worked on practical implementation of the Vivid Agents model, the open source language Prova (www.prova.ws), that is still actively developed. For the last five years, he has been a Senior Researcher at Betfair, focusing on event-driven distributed systems, in particular, the in-house Event Processing Network, serving as a W3C representative, working in the Rule Interchange Format group, serving as a member of the Steering Committee of RuleML, co-authoring the Reactive RuleML specification, contributing to Event Processing Technical Society, and co-chairing the industrial track of the upcoming DEBS 2011 conference in Yorktown Heights.