Availability Assessment of a Cloud Server System: Comparing Markov and Semi-Markov Models

Nowadays Markov Modelling Processes (MMP) found widely use into realm pertaining to availability assessments of diverse complex technical systems and devices. In this regard, Cloud Server Systems (CSSs) are no exception. However, in order to leverage MMPs for the availability evaluation of CSSs the researchers should take into account that this assessment procedure based on the utilization of just one distribution, that is exponential distribution of times to failure, times to repair etc. It is a serious constraint, which would be need preliminary substantiate considering different negative events and impacts. In that regard, to note that some groups of researchers into realm pertaining to CSSs availability are increasingly trying to leverage approaches based on Semi-Markov Modelling Processes (SMMPs). Using SMMPs the researchers can consider more number of parametric characteristics and have a realistic picture as regards change of availability level for the server system. In addition, seeking to improve overall availability of CSSs, the researchers can determine accuracy of numerical modeling results for SMMPs in the so-called borders' states, when Non-Markovian Process transforms at Markovian Process. By considering significance of these aspects, authors of this paper propose to perform mathematical modeling of behavior of a cloud server system based on joint usage of MMPs and SMMPs and get availability assessments considering diverse influence factors on states of the concrete CSS. This information is to be studded and can be used by developers in order to as well as design of the CSSs considering different influence factors.