Authors: Sascha Bosse, Matthias Splieth, Klaus Turowski
In today's world, more and more systems are supported by IT services. High availability of these IT services is often crucial to customers and ? in order to avoid penalty costs for service-level violation ? also for providers. Decisions affecting IT service availability are mainly made in the service design stage and may be costly to correct afterwards. Model-based prediction models can support decision making in this stage. However, the suitability of the existing approaches is still discussed in literature. In this paper, a quantitative evaluation of the state of the art in IT service availability prediction is performed based on a structured literature review. Results indicate that no standard approaches do exist to this moment in time. One reason for this fact is the limited comparability of these models due to the lack of quantitative and comparable evaluations.