Energy-efficient Virtual Machine Scheduling In Cloud Computing

The emergence of cloud computing has revolutionized the informationtechnology industry. Based on the pay-as-you-go model, it enables elasticon-demand provisioning of computing resource. As the proliferation of cloudcomputing, large-scale data centers have been established around the word, whichcontains thousands of compute nodes. However, data centers hosting cloudapplications consume huge amounts of energy, contributing to high operating costsand carbon footprints to the environment.By applying dynamic consolidation of Virtual Machines (VMs) in cloud datacenters, we can efficiently improve the utilization of computing resource and reduceenergy consumption. However, the server may overload during the VMsconsolidation which will affect the server’s performance and the service quality.To reduce energy consumption of cloud data center, this thesis presents adistributed heuristic energy-efficient Virtual Machines (VMs) scheduling algorithmin cloud to improve the computing resource utilization under qualit y of serviceconstraints, thus reducing the energy consumption and carbon emissions. Moreover,this algorithm keeps the service level agreement violation rate in a low level toguarantee the service quality. The main contributions are as follows:Firstly, the algorithm has four steps: host overloaded detecting, hostunder-loaded detecting, VM selection and host selection. By analyzeing the energyconsumption of VM dynamic migrations when host is overloaded, we haveproposed a heuristic strategy.Secondly, we use this strategy and VM load prediction to determine whetherthe host is overloaded or not. And thus, we can consolidate the VMs to reduceenergy.Finally, we have simulated a large-scale data center using workload traces frommore than a thousand PlanetLab VMs in CloudSim platform. The result shows thatthe proposed algorithm significantly reduced the energy consumption and the levelof SLA violations.