Scalable fault tolerant architecture for complex event processing systems

Abstract

Complex Event Processing (CEP) is one of the emerging areas in computer science and it is being heavily used in real time systems during the recent times. CEP systems should be scalable and fault tolerant. The research project epZilla is to build a scalable, highly available and fault-tolerant distributed architecture for CEP systems. Software Transactional Memory, Leader Election algorithms, dynamic load balancing, dynamic service discovery, Stratification are the concepts which are used to build the proposed architecture. The results of the pilot run are promising and show that the proposed architecture is scalable and fault tolerant.