Performability models for designing disaster tolerant Infrastructure-as-a-Service cloud computing systems

Providing high availability in a diverse client demand scenario is a key challenge of cloud computing providers. A possible approach to protect cloud systems from natural disasters corresponds to the utilization of redundant data centers located far enough apart. The time to migrate a virtual machine (VM) from a data center to another increases due to distance. As this time impacts the overall system behavior, performability evaluation considering VM migration time and different load balances is of utmost importance when considering the analysis of distributed cloud systems. This work presents performability models for evaluating distributed cloud computing systems deployed into multiple data centers considering disaster occurrence. Additionally, we present a case study which evaluates a set of scenarios with different client loads and distances between data centers.