Parameter optimization for hybrid auto-scaling mechanism

Elastic resource scaling is a key feature of cloud computing. The most popular approach to managing elasticity is scale-out, but this approach sometime incurs performance problems, especially for service with rapid workload change. So we proposed the hybrid auto-scaling method which use both scale-out and scale-up as workload pattern. However, the effect of the method depends on the parameters. Moreover, the required service performance and the budget depend on a SLA that service providers conclude with their customers. In this paper, we present the parameter optimization method to keep a certain service level and a budget. We then evaluate the effect of the method and show it can adjust the parameters for the experimental workload.