EStream: An Integration Of Event And Stream Processing

Abstract

Event and stream data processing models have been researched extensively and are utilized in diverse application domains. Advanced applications require both event and stream processing, which is currently not supported. Although there are number of similarities and differences between them, a synergistic integration of their strengths will be better than the sum of their parts. In this thesis, we present EStream, an integrated event and stream processing system for monitoring changes on stream computations and for expressing and processing complex events on continuous queries (CQs). We introduce attribute-based constraints for reducing uninteresting events that are generated from CQs. We discuss the generalized specification of CQs, complex events, and rules. We also discuss stream modifiers, a special class of stream operators for computing changes over stream data.