GMPR: A Two-Phase Heuristic Algorithm for Virtual Machine Placement in Large-Scale Cloud Data Centers

Abstract

In the cloud data centers, due to the variable resource requirements of cloud tenants, designers applied the infrastructure as a service (IaaS) model to provide services for tenants with allocating in charging. The application of virtualization technology enables multiple virtual machines (VMs) to share the resources of a physical machine (PM). Meanwhile, the efficiency of the data centers greatly depends on the working efficiency of VMs. Virtual machine placement (VMP) plays a vital role in minimizing total energy consumption and resource wastage for cloud data centers (CDCs). In this article, we propose a greedy algorithm minimizing power consumption and resource wastage (GMPR) for the VMP scheme to address the abovementioned issues. GMPR prioritizes the power-efficiency of PM to reduce the number of active PMs and to minimize total energy consumption. In addition, reducing total resource wastage involves first minimizing the resource of balance and resource wastage for a novel PM hosts VM and second minimizing resource wastage for PM placed VM. Extensive simulation results are conducted on synthetic and instances of Amazon EC2 with performance metrics confirm that GMPR has superiority in reducing energy consumption and resource wastage by an average of 1.91% and 16.18% compared with a cutting-edge method.