Timing behavior anomaly detection in enterprise information systems

Business-critical enterprise information systems (EIS) have to satisfy high availability requirements. In order to achieve the required availability, automatic failure detection and diagnosis techniques must be used. A major cause of failures in EIS are software faults in the application layer. In this paper, we propose to use anomaly detection to diagnose failures in the application layer of EIS. Anomaly detection aims to identify unusual system behavior in monitoring data. These anomalies can be valuable indicators for availability or security problems, and support failure diagnosis. In this paper we outline the basic principles of anomaly detection, present the state of the art, and typical application challenges. We outline a new approach for anomaly detection in Enterprise Information Systems that addresses some of these challenges.