A software availability model based on multilevel software rejuvenation and markov chain

Abstract

Increasing use of software, rapid and unavoidable changes in the operational environment bring many problemsfor software engineers. One of these problems is the aging and degradation of software performance. Software rejuvenationis a proactive and preventive approach to counteract software aging. Generally, when software is initiated, amounts ofmemory are allocated. Then, the body of software is executed for providing a service and when the software is terminated,the allocated memory is released. In this paper, a rejuvenation model based on multilevel software rejuvenation andMarkov chain presented. In this model, the system performance as a result of degraded physical memory and memoryusage is divided into four equal levels by services. Hence, we offer four types of policies for software rejuvenation. Inaddition, the system availability is determined, and a cost function for the model is introduced. The cost functionincludes the time of performing rejuvenation, the number of system services at any time, and the number of rejuvenationactions. To validate the proposed model, a case study in the banking system in Iran has been studied. Due to thedifferences in the use of the system over time, it is better to perform the four different policies with regard to the useof the system. The numerical results show that the proposed model is convenient for the system so that the costs arereduced per day.