Creating a knowledge database on system dependability and resilience

The paper deals with the problem of creating a knowledge database on system dependability and resilience, created on the basis of available system and application logs. Special tools to collect and analyse these data from many systems have been developed. Taking into account a wide spectrum of various logs we explore them locally and globally. This allowed for identification of characteristics of normal operation and anomalous behaviour. A lot of attention is paid to the problem of selecting measures to identify symptoms characterising system operation and their usefulness in dependability and resilience evaluation or prediction. The concepts presented are illustrated with experience gained during monitoring of real systems.