A Stochastic Reward Net Approach to model availability of cloud virtualization

Cloud computing initiates various opportunities and challenges for information technology industry. Due to distributed and virtualized environment of cloud, component failures are quite common phenomena. These failures usually lead to sporadic downtime and degrade the availability level of cloud services. Availability of service is stipulated in the Service Level Agreement (SLA) and calculated as the percentage of time in which the service will be up over a year. This paper deals with the availability analysis of cloud using Stochastic Reward Net (SRN) approach. We consider the failure of active and reserved servers along with Virtual Machines (VMs) and Virtual Machine Monitors (VMMs) into account. Finally, the SRN model applied into the Stochastic Petri Net Package (SPNP) and numerical results are presented. Furthermore, the proposed model has captured independent and dependent failures.