数据中心虚拟机带宽优化策略研究

发布时间：2018-02-22 00:41

本文关键词： Openstack 数据中心虚拟机流量预测带宽优化　出处：《西南交通大学》2017年硕士论文　论文类型：学位论文

【摘要】：随着信息社会的高速发展,互联网行业对资源合理分配的要求越来越高,为满足上述要求,数据中心应运而生并发展成为了互联网行业基础设施的核心。数据中心资源分配一直是研究的热点。当前云数据中心内,虚拟机流量种类的丰富多样,规模的大小不一,已经成为了数据中心性能提升的瓶颈,使得数据中心带宽资源的合理利用变得复杂和低效,急需合理的带宽资源利用方案,以提高带宽资源的利用率。本文对数据中心带宽利用率低的问题进行了研究,以数据中心虚拟机历史流量特征为依据对虚拟机未来流量特征进行预测,实时得出虚拟机流量使用趋势,同时结合动态带宽调节策略,减少虚拟机因预测算法的迟滞性带来的影响。在动态调节预测值与实时流量的过程中加入对最小带宽保证策略,并以上述预测及优化结果来制定相应的带宽策略,达到提升数据中心带宽利用率的目的。本文主要研究工作如下:1.调研现有数据中心平台,采用Openstack云平台架构为实验框架,并分析了其网络架构特点及其功能。研究了虚拟机网络通信的机制与原理,并搭建了数据中心主要实验环境。设计并开发了网络组件Neutron部分功能,根据虚拟机网络通信原理与机制,实现了基于linux系统底层Iptables规则虚拟机带宽的监测与控制。2.依据虚拟机历史流量,将遗传算法优化的BP神经网络算法应用于虚拟机网络带宽预测并得到较好的预测结果。为了解决虚拟机带宽预测机制在特定流量特征下产生迟滞性的问题,将预测算法与动态带宽调节算法相结合,优化虚拟机预测带宽,并实现最小带宽保证。3.将本文提出的带宽优化策略应用于Openstack框架下搭建的数据中心。通过模拟虚拟机不同流量情况,将本文策略与虚拟机静态带宽预留策略相比较并进行了实验验证。实验结果证明,本文虚拟机带宽优化策略具有明显优势。综上所述,本文依据虚拟机历史流量数据的特征,提出了虚拟机带宽优化策略,并将其在数据中心进行了实验测试。实验结果表明,本文提出的虚拟机带宽策略能够有效提高数据中心内虚拟机网络带宽利用率。
[Abstract]:With the rapid development of the information society, the requirements of the Internet industry for the rational allocation of resources are becoming higher and higher, in order to meet the above requirements, Data center has emerged as the times require and developed into the core of the Internet industry infrastructure. Resource allocation of data center has been a hot topic all the time. At present, in cloud data center, virtual machine traffic is rich and diverse, and the size of virtual machine is different. It has become the bottleneck of improving the performance of data center, which makes the reasonable utilization of bandwidth resource become complex and inefficient. In order to improve the utilization of bandwidth resource, this paper studies the problem of low bandwidth utilization in data center, and predicts the future traffic characteristics of virtual machine based on the historical traffic characteristics of virtual machine in data center. The trend of virtual machine traffic is obtained in real time, and the dynamic bandwidth adjustment strategy is combined to reduce the influence of the hysteresis of the prediction algorithm. The minimum bandwidth guarantee strategy is added in the process of dynamically adjusting the predicted value and the real-time traffic. Using the above prediction and optimization results to formulate the corresponding bandwidth strategy to improve the bandwidth utilization of the data center. The main research work of this paper is as follows: 1. The existing data center platform is investigated, and the Openstack cloud platform is used as the experimental framework. The characteristics and functions of the network architecture are analyzed, the mechanism and principle of virtual machine network communication are studied, the main experimental environment of the data center is built, and some functions of the network component Neutron are designed and developed. According to the principle and mechanism of virtual machine network communication, this paper realizes the monitoring and control of virtual machine bandwidth based on linux system bottom Iptables rule virtual machine bandwidth. The BP neural network algorithm optimized by genetic algorithm is applied to the bandwidth prediction of virtual machine network and a good prediction result is obtained. In order to solve the problem that the bandwidth prediction mechanism of virtual machine produces hysteresis under specific traffic characteristics. Combining the prediction algorithm with the dynamic bandwidth adjustment algorithm, the prediction bandwidth of virtual machine is optimized. The bandwidth optimization strategy proposed in this paper is applied to the data center built under the framework of Openstack. Compared with the static bandwidth reservation strategy of virtual machine, the experimental results show that the bandwidth optimization strategy of this paper has obvious advantages. In summary, according to the characteristics of virtual machine historical traffic data, The virtual machine bandwidth optimization strategy is proposed and tested in the data center. The experimental results show that the proposed virtual machine bandwidth strategy can effectively improve the virtual machine network bandwidth utilization in the data center.
【学位授予单位】：西南交通大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP308;TP302

【相似文献】