Events and streams: harnessing and unleashing their synergy!

• Authors: Sharma Chakravarthy;Raman Adaikkalavan
• Venue: Proceedings of the second international conference on Distributed event-based systems
• Year: 2008

One of the purposes of this paper is to demonstrate the ubiquitous nature and relevance of simple and complex events in situation monitoring and other unforeseen applications. This paper retrospectively analyzes the progress of event-based capability and their applicability in various domains. Although research on event-based approaches started in a humble manner with the intention of introducing triggers in database management systems for monitoring application state and to automate applications by reducing/eliminating user intervention, currently it has become a force to reckon with as it finds use in many diverse domains. This is primarily due to the fact that a large number of real-world applications are indeed event-driven and hence the paradigm is apposite. The other purpose of this paper is to identify and establish the synergy between stream processing and event processing. The resurgence of complex event processing (or CEP) has demonstrated the resiliency of this paradigm and its relevance to a large class of applications. In addition, the advent of stream processing based on sensor and other data generated on a continuous basis has, in our opinion, enhanced the role of events in critical ways. Instead of implicitly assuming event generation, stream processing indeed provides a powerful mechanism for generating interesting events. In this paper, we briefly overview the development of the ECA (or event-condition-action) paradigm. We briefly discuss the evolution of the ECA paradigm (or active capability) in Relational and Object-oriented systems. We then describe several diverse applications where the ECA paradigm has been used effectively. The applications range from customized monitoring of web pages to specification and enforcement of role-based access control policies (RBAC). The multitude of applications clearly demonstrate the ubiquitous nature of event-based approaches to problems that were not envisioned as the ones where the active capability would be applicable. Finally, this paper analyzes the differences between stream and event processing and proposes an integration architecture to meet the requirements of applications such as linear road bench mark and network fault management applications. These applications have a strong stream computation component to generate interesting events and a complex event processing component to detect situations for timely notification.