Today, data centers are the main source of providing cloud services through a service level agreement (SLA). Most research papers for cloud resource management concentrate on how to reduce host energy consumption and SLA violation (SLAV) to minimize operational cost. However, they do not consider the amount of penalty that cloud provider should pay to users because of SLAV. In this paper, we propose a new penalty-aware and cost-efficient method that considers cloud resource management as a cost problem. In this method parameters such as user budget, penalty, and host energy consumption cost play an important role in minimizing operational cost which leads to higher profit for cloud provider. The simulation results with CloudSim show that our proposed method minimizes operational cost compared to the prior resource managements.
INTRODUCTIONCloud computing is becoming more and more popular as an easily usable alternative service deployment platform [1]. Rapid increase in demands for cloud services has forced major cloud service providers such as IBM, Amazon, Google, etc. to build more and more data centers. Large data centers, as the main source of providing cloud services, are composed of hundreds of computational hosts, tens of high speed switches and other network devices that consume a significant amount of energy. Hardware infrastructures such as servers, storage devices and network elements consume significant amount of energy when either they are idle or not fully utilized [2]. Hence, data center energy efficiency can be effective in decreasing operational costs. As operational costs decrease, cloud service provider profit increases. Moreover, because higher energy consumption causes more carbon dioxide (CO 2 ) emission [3], efficient energy consumption policies are expected to help reducing CO 2 emission. Hence, an energy efficient resource management is a must.Various effective methods are proposed for reducing the power consumption of devices in data centers, for example server consolidation using virtualization [4,5]. Virtualization technology provides a facility to run several isolated applications by creating a virtual machine (VM) for each application on a host. This leads to increase in host utilization and recently all data centers are using virtualization for providing cloud services [2]. More precisely, virtualization technology (e.g. Xen