Research On Virtual Machine Placement Problem Based On Multi-objective Optimization

A reasonable virtual machine(VM)allocation strategy is one of the key technologies to improve the performance of cloud computing systems and reduce energy consumption.The rapid development of cloud computing technology enables users to acquire and use computer resources,and not limited by time and geography.However,large-scale cloud data centers have meet the problems,such as energy consumption and big pollution.To solve these problems,this paper proposes solutions in the initial placement phase and dynamic management phase of the VM respectively.The initial of VM placement significantly influences the utilization of data center resources and energy loss.This paper proposes a VM initialization placement strategy based on energy consumption and resource consumption for optimization.Through the analysis and research of related theories,this paper adopts a Genetic Algorithm(GA)with global search,parallel computing,better robustness and stability,and easy to achieve multi-objective optimization.Experiments show that compared with the traditional heuristic method,the GA proposed in this paper can effectively improve the utilization of server resources and save energy consumption.The phase of dynamic management of VM placement mainly refers to the process in which the VM needs to be relocated due to the dynamic change of cloud data center load.In this stage,this paper mainly considers the stability of VM placement.The optimization goal is to use both the distribution stability of multiple VMs and the number of VM migrations using classical Algorithm NSGA?.Experiments show that compared with the traditional optimization methods,the proposed method can well balance the stability of the physical nodes and the number of migration of VMs,improve system resource utilization,and reduce the cost of VMs migration.