An adaptive virtual machine placement Gaussian model and method

Multiple types of resources in data centers are mutually constrained. Once a physical resource of hosts has a large overload risk, the hosts cannot be deployed the living migration VMs even if there are any remaining resources, resulting in waste of resources. To improve the comprehensive resource utilization, a virtual machine placement scheme and method based on adaptive Gaussian model (AGM-VMP) is proposed. First, the AGM-VMP method declares that when multiple resources are in the normal state at the same time in data centers, it facilitates reducing the waste of resources and degrades energy consumption. Next, the adaptive Gaussian model is developed to estimate the normal workload probability of hosts. Finally, the model, scheme and method of virtual machine placement is treated according to the presented normal workload probability of hosts. Experimental results show that the AGM-VMP method improves energy utilization, quality of service and comprehensive resource utilization.