Semi-Markov Availability Model for Infrastructure as a Service Cloud Considering Energy Performance

This paper addresses a stochastic approach based on Semi-Markov Process modeling in order to determine availability level and energy performance for Infrastructure as a Service (IaaS) Cloud. It is generally agreed today that Cloud Service Providers try to offer diverse IT services, using large number of physical machines (PMs). It really is created serious energy performance issue for cloud computing systems. In this situation we also propose to use information of Technical and Information State Control System in order to improve monitoring process for IaaS Cloud. Therefore, main part of the efforts has been devoted to the development of a monolithic Semi-Markov Availability model for IaaS Cloud and the evaluation power consumption for non-failed, that is available PMs. Semi-Markov Availability model for the cloud infrastructure considering sudden failures of physical and virtual machines are considered, too. Mathematical modeling results will be present in order to give certain clarifications on the possible use of approach suggested.