Dynamic Resource Configuration Research Based On Multi-objective Optimization In Cloud Computing

With the help of system virtualization technology and virtual machine(VM)real-time migration technology,cloud computing realizes dynamic on-demand configuration of computing resources,which greatly improves the efficient use of resources.How to balance the relationship between the stability of physical nodes and energy consumption in the process of resource allocation is a basic and complex problem.However,most of the existing cloud platform resource allocation strategies implement green computing by reducing the number of active physical nodes,and these resource scheduling schemes do not consider the stability of physical nodes according to the dynamic nature of the application load.At the same time,it is easy to obtain the local optimal solution by considering only the stability of the resource distribution mode under the current state or the optimization of a single objective.This paper considers several factors such as the stabilization time of the cloud platform,the active physical nodes and the number of VM migrations and based on prediction information of applications' workloads,VM distribution on nodes was defined as a multiobjective optimization problem and the multi-objective genetic algorithm based on improved NSGA-II(MOGAINS)was employed to find the approximate optimal solution of that problem.The simulation results show that MOGAINS uses the type-matching rule(TMR)to realize the balanced utilization of resources,and increases the distribution of the evolutionary population through the pre-screening operation of the Pareto front.The effectiveness of the genetic algorithm is verified by the analysis of the evolution process.After many experiments,it is found that a large number of individuals are distributed in the same Pareto front in MOGAINS,and the density information around the individual can't be well reflected by the same level of non-dominated sets of individuals.In the following,based on MOGAINS,the multi-objective genetic algorithm based on improved SPEA-II(MOGAISP)was used for the best resource scheduling.The simulation results show that the performance of the MOGAINS is slightly better than that of the MOGAINS.Compared with the MOGAINS,the MOGAISP can ensure the stability of the physical node,so that the new VM distribution mode has lower power consumption,the benefit of the cloud service provider is greatly improved,and under the background of the increasing demand of the cloud computing,this paper has some practical significance and economic value.