An Evaluation of a Complex Event Processing Engine

Complex event processing (CEP) is a technology for capturing what has happened in the real world from a huge amount of sensor data. This paper evaluates a CEP engine Esper to find an efficient method for CEP. The authors are trying to clarify the factors that affect the throughput of Esper. They investigated the effects on the throughput of the three factors: the input rate, the window size of the algorithm for processing events, and the length of a pattern for matching. The results of the evaluation show that the throughput of Esper is extremely decreased for some algorithms with a large window size and matching with a long pattern.

I. Introduction

With the spread of computers and sensors, many kinds of and a large amount of data became available for capturing what has happened in the real world. The something that has happened is called an event. Event processing [1] means computational operations on events, and we also call a set of data that represents an event in the real world an “event”. To extract useful knowledge from the huge amount of data, complex event processing (CEP) [1] has been receiving attention. Developing an efficient method of CEP is expected to innovate in event processing.