Energy Efficient Virtual Machine Placement Approach for Balanced Resource Utilization in Cloud Environment

Abstract

Cloud is a business model that provides on demand computing resources to the clients on rental basis. It is an utility model, so clients need to pay only for that amount of resources which is actually used. Cloud provider want to gain more and more profit that can be possible only by effective resource utilization. For this purpose virtualization technology is used, which allows sharing of physical resources. In virtualization hypervisor provide requested resources to the client in the form of virtual machine. Virtualization increases the resource utilization by creating the number of VM in a single host, but wrong VM placement can increase resource leakage. Resource leakage can be defined as an amount of waste resources that can’t be used. Resources in the cloud are in multi dimensions i.e. CPU, RAM, bandwidth etc. So resource leakage is one of the major problems in the cloud environment. Balanced resource utilization in the cloud is more challenging task as compared to the traditional distributed computing paradigms, due to dynamic behavior of the VM resource requirement. Number of virtual machine placement algorithms has been proposed that run under cloud computing environment. Main goal of these algorithms are either to save energy by shutting down some severs or maximizing the resources utilization, but few of them considered both. This paper proposes a VM placement approach that increases the balanced resources utilization and reduced power consumed by the data centers. For the balanced resources utilization six clusters CMB, CBM, MBC, MCB, BMC and BCM are used, where C, M and B stands for CPU, memory and bandwidth respectively. Initially some host assigned to each cluster according to the remaining capacity. CBM cluster keep the host where remaining CPU > remaining bandwidth > remaining RAM. Host in the CBM cluster can run only that VM, which require more CPU and minimum RAM. This approach is motivated by the concept that VM is assigned to PM such that, resource requirement of the VM are complement with the resource utilization of the PM. CloudSim simulator is used to implement this approach. Experimental result shows that this approach increased the resource utilization, reduced the resource leakage and reduced the energy consumption by minimizing the number of active server.