Stochastic model of performance and cost for auto-scaling planning in public cloud

Cloud computing has the potential to reduce the cost of systems on the Internet. Elasticity mechanisms, such as auto-scaling, enable avoiding wastes, delivering only the necessary resources. Defining and implementing an efficient auto-scaling policy is a complex task that depends on the parameter setting, types of VM contracts and the expected workload. All those variables must be taken into account when establishing the tradeoffs between performance and cost to fulfill a given servicelevel agreement (SLA). We propose a stochastic model to assist in cloud planning. The model was validated for a set of significant scenarios by comparing the respective model's results with those obtained from real system measurements. This model takes as input the auto-scaling configuration parameters and the time between user requests. The proposed model is employed to compute throughput, mean response time, and cost of the cloud computing infrastructure setup. A sensitivity analysis was also conducted for identifying the parameters impact on the system performance.