基于SDN的数据中心网络流量调度方法研究

发布时间：2018-02-27 15:23

本文关键词： SDN OpenFlow 数据中心网络老鼠流大象流　出处：《湘潭大学》2017年硕士论文　论文类型：学位论文

【摘要】：随着社交网络、移动互联网、物联网等业务领域的发展,数据中心作为这些业务所依赖的基础设施,已经发生了许多重大的变化。数据中心网络规模的不断增大,网络业务呈现多样化、复杂化,使得网络流量快速增长,因此数据中心网络流量的调度成为了一个研究热点。由于传统网络静态的运作模式和僵硬的管理方法,不能灵活、快速、智能的调度底层网络流量,使得数据中心内的流量调度研究发展缓慢。SDN作为新兴的网络架构,实现了网络控制与转发的分离,利用集中的方式来实现网络的转发功能以及细粒度的流量控制能力,使得SDN技术在数据中心网络中有着广泛的应用和部署。本文通过将SDN技术应用到数据中心网络中,提出了一种基于SDN的实时流量调度方法(Real-time Traffic Scheduling Method based on SDN,RTSM-SDN),使得流量调度更加灵活智能,从而提高了网络资源利用率。本文的主要工作和创新如下。(1)分析了数据中心内流量调度相关研究背景和现状,传统的等价多路径流量调度算法已经不能满足需求。结合流量调度的实现原理,将SDN技术应用到数据中心网络中。在SDN网络架构下,通过控制器与交换机之间OpenFlow消息的交互,可以获得底层网络的性能参数(当前链路可用带宽、链路时延、丢包率等)。在计算路由时将这些实时获取的参数加入到路由计算度量中,这样在计算路径时既考虑了网络拓扑,同时也考虑了网络当前状况,尽可能避免了网络拥塞,提高了网络服务质量。(2)结合数据中心网络流量特点,根据大象流与老鼠流的传输需求,将获取的实时链路时延和丢包率作为老鼠流传输的路由计算度量,将链路当前可用带宽作为大象流传输的路由计算度量,在此基础上再加入传输时延的硬阈值作为限制条件,建立目标函数,即最小化网络的路由度量。实验表明,在传输时延、网络整体吞吐量以及数据包重传次数三个性能指标下,RTSM-SDN方案与ECMP、Hedera、RepFlow相比,能在保障老鼠流传输的同时,也很好地保障了大象流的传输,大大增加了网络吞吐量和链路利用率。
[Abstract]:With the development of social network, mobile Internet, Internet of things and so on, data center, as the infrastructure on which these services depend, has undergone many important changes. The network traffic is diversified and complicated, which makes the network traffic grow rapidly. Therefore, the scheduling of network traffic in data center becomes a research hotspot. Because of the static operation mode and rigid management method of traditional network, it is not flexible. The rapid and intelligent scheduling of the underlying network traffic makes the research of traffic scheduling in the data center develop slowly. SDN as a new network architecture realizes the separation of network control and forwarding. Using centralized method to realize network forwarding function and fine-grained flow control ability, SDN technology has been widely used and deployed in data center network. This paper applies SDN technology to data center network. This paper presents a real-time traffic scheduling method based on SDN, which makes real-time Traffic Scheduling Method based on SDN RTSM-SDN (RTSM-SDN) more flexible and intelligent. The main work and innovation of this paper are as follows: 1) the research background and current situation of traffic scheduling in data center are analyzed. The traditional equivalent multi-channel runoff scheduling algorithm can no longer meet the demand. Combined with the realization principle of traffic scheduling, the SDN technology is applied to the data center network. Under the SDN network architecture, the exchange of OpenFlow messages between the controller and the switch is achieved. The performance parameters of the underlying network (current link available bandwidth, link delay, packet loss rate, etc.) can be obtained. At the same time, it also considers the current situation of the network, avoids the network congestion as far as possible, improves the network service quality and combines the characteristics of the network traffic in the data center, according to the transmission requirements of the elephant flow and the mouse flow. The obtained real-time link delay and packet loss rate are used as the routing calculation metrics for rat stream transmission, and the current available link bandwidth is used as the routing calculation metric for elephant stream transmission. On this basis, the hard threshold of transmission delay is added as the limiting condition. The objective function is to minimize the routing metric of the network. Experiments show that RTSM-SDN scheme can guarantee rat stream transmission at the same time as ECMPN Hedera RepFlow under three performance indexes: transmission delay, overall network throughput and packet retransmission times. It also ensures the transmission of elephant stream and greatly increases network throughput and link utilization.
【学位授予单位】：湘潭大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP393.0

【参考文献】