Cost-Aware Automatic Virtual Machine Scaling in Fine Granularity for Cloud Applications

It is a tendency for enterprises to deploy their applications on Infrastructure as a Service (IaaS) platforms. Many latest IaaS service providers offer Virtual Machine (VM) instances with various capacities and prices by the minute. In this paper, based on the observation that the workload of nowaday applications fluctuates frequently, we propose a cost-aware automatic VM scaling method of fine granularity to satisfy the Service Level Agreement (SLA) and minimize the rent of VMs down to the minute. In the environment with sporadic and sharp swings of workload, our approach acts quickly to get suitable VM scaling scheme to stabilize the response time, reduce the SLA violations and save the rent of VM usage.