Predicting an IT Service’s Availability with Respect to Operator Errors

Ensuring high availability of IT services is crucial for IT service providers, not least because of the cloud computing para-digm. Thus, the management of IT system landscapes should be supported by availability predictions. Although operator errors account for a high number of service downtimes, only few approaches for quantitative availability prediction consider operator actions and errors. Also mechanisms that were developed to face the high number of operator errors cannot be modeled with existing approaches. Therefore, a new design for availability prediction of IT services is developed. Based on a flexible petri nets’ simulation, operator errors and mechanisms to face them are introduced. An evaluation shows that the developed design is correct and results are plausible.