Reduction of Power and Virtual Machine Migration Inside a Cloud Datacenter

Abstract

Today’s datacenter consumes high amount of electrical energy for its operation. It increases day by day, thus increasing the operation cost and carbon dioxide emission. To reduce the power consumption at a datacenter, a model with an adaptive threshold virtual machine (VM) consolidation method that reduces the number of VM migration and power consumption in the datacenter was used. By consolidating VMs and switching off unused hosts, a cloud provider can reduce physical resource usage and power consumption. But due to service level agreement(SLA) between a cloud service provider and cloud users, a cloud service provider cannot degrade the performance. To balance between energy and performance, a cloud service provider must maintain a reasonable performance while reducing power consumption. In this thesis work a novel technique have been proposed to adhere both power and performance by implementing an adaptive host upper utilization threshold and host lower utilization threshold for CPU utilization in a host, this makes our model more robust than the existing models. Our proposed algorithm reduces the power consumption and limits the number of VM migration while ensuring high level of SLA. To simulate the experiment and to validate our proposed algorithm using real world datacenter datasets, we have used CloudSim which is a widely popular datacenter simulation toolkit.