Exploring the space of system monitoring

Abstract

During system exploitation and maintenance an important issue is to evaluate its operational profile and detect occurring anomalies or situations which may lead to such anomalies (anomaly prediction issue). To resolve these problems we have studied the capabilities of standard event and performance logs which are available in computer systems. In particular we have concentrated on checking the morphology and information contents of various event logs (system, application levels) as well as the correlation of performance logs with operational profiles. This analysis has been supported with some tools and special scripts. © 2013 Springer-Verlag Berlin Heidelberg.