Research On Optimization Strategy Og Data Center Virtual Machine Bandwidth

With the rapid development of the information society,the Internet industry is increasingly demanding the allocation of resources.The data center has emerged and developed into the core of the Internet industry infrastructure,with the development of the rhythm of the gradual acceleration of its efforts to meet the model gradually changed to maximize the use of resources.The resource allocation of data center has always been the focus of research,especially the data resources of the data center,the variety of traffic types,the size of the flow of different sizes,making the rational use of bandwidth resources become complex and inefficient.Data center bandwidth resources have become the bottleneck of data center performance improvement,the need for reasonable use of bandwidth resources to improve the utilization of bandwidth resources.Based on the historical traffic characteristics of the data center virtual machine,this paper predicts the future traffic characteristics of the virtual machine,realizes the trend of virtual machine traffic usage in real time,and combines the dynamic bandwidth adjustment strategy to reduce the hysteresis of the virtual machine due to the prediction time Impact.The minimum bandwidth assurance strategy is added to dynamically adjust the predicted and real-time traffic,and the corresponding bandwidth strategy is developed with the final result.The main work of this paper is as follows:1.Research the existing data center platform,using Openstack cloud platform architecture to build data center experimental architecture,and analysis of its network architecture features and functions.This paper studies the mechanism and principle of virtual machine network communication,and constructs the main experimental environment of data center.It designs and develops the function of Neutron part of network component.According to the principle and mechanism of virtual machine network communication,the virtual machine bandwidth based on linux system bottom layer Iptables is realized Monitoring and control.2.Based on the historical traffic of virtual machine,the BP neural network algorithm optimized by genetic algorithm is applied to the prediction of virtual machine network bandwidth and the prediction result is obtained.In order to solve the hysteresis of this kind of virtual machine bandwidth prediction,the prediction algorithm and the dynamic bandwidth adjustment algorithm are combined to optimize the virtual machine to forecast the bandwidth and realize the minimum Bandwidth guarantee.3.The bandwidth optimization strategy proposed in this paper is applied to the data center under the Openstack framework.By simulating the different traffic conditions of the virtual machine and comparing the virtual machine bandwidth utilization with the virtual machine static bandwidth reservation strategy.The results show that the virtual machine bandwidth optimization strategy has obvious advantages.In summary,this paper based on the characteristics of virtual machine historical traffic data,proposed virtual machine bandwidth optimization strategy,and in the data center to achieve.Through the simulation of virtual machine traffic experiment,the proposed virtual machine bandwidth strategy effectively improves the virtual machine subnet bandwidth utilization.