Profit-Aware DVFS Enabled Resource Management of IaaS Cloud
Shahzad Ali, Siyuan Jing · 2013
Power consumption of IaaS Cloud is growing at a rapid pace during the last decade. This leads to high operational and maintenance cost, and also reducing the profit margin of IaaS Cloud providers. Apparently profit margin depends on operational cost and revenue earned to comply with the Service Level Agreement (SLA) on the basis of achieved performance level. In this paper, we are building the relationship between energy consumption of servers and fulfillment of SLA (QoS) in IaaS Cloud. So we propose a framework to maximize revenues from SLA achieved requests on the energy efficient resource management in IaaS Cloud, in which the servers are equipped with DVFS module. Servers are not allowed to be turned on or turned off. Moreover, we consider a common cost of virtual machine (VM) migration or start/stop. This framework explains four sub-problems, i.e. service placement which encapsulating in VM, resource capacity allocation, load balancing and dynamic voltage/frequency scaling. A hybrid optimization solution is proposed to solve these problems and its effectiveness, compared to existing approach, is evaluated using synthetic workloads. Results show that our solution as compared to others can yield significant profit gain for the cloud provider.