面向故障恢复的域内流量工程技术研究与实现

发布时间：2018-10-31 19:38

【摘要】：随着互联网快速发展,网络应用层出不穷,网络承载的流量及业务越来越多,对网络的可靠性提出了巨大挑战。但网络故障时有发生,严重影响了网络的正常运行,给用户及网络服务商带来巨大损失。研究者对网络故障恢复做了一些研究,主要是基于连通性角度进行网络故障恢复,将流量从故障路径上迁移到恢复路径上来,而没有考虑到网络恢复后的负载均衡,很容易造成新的网络拥塞;而对于流量工程的研究能够做到提高网络资源利用率,减小网络链路最大利用率,使网络流量均衡分布到网络中,但是一般采用预规划的方式对一段时间内的流量进行最优规划,对于网络故障的实时响应不足。随着网络体系结构的研究发展,SDN概念的提出给我们指出了新的思路,我们基于SDN集中式的路由体系结构,提出了面向故障恢复的域内流量工程系统,并提出了相应的在线优化算法MCPF,针对实际网络运行情况对网络中流请求进行规划,使网络达到负载均衡,并且当网络中发生故障时能够迅速针对故障进行路径优化,快速恢复故障并确保故障后的流量负载均衡。本论文的主要工作如下:第一,提出了一种面向故障恢复的域内流量工程系统(FR-TE)结构。针对故障条件下流量工程的优化需求,基于集中控制的思想,提出了面向故障恢复的域内流量工程系统结构。该系统结构通过对网络转发节点上的流量请求进行实时规划部署,使网络流量均衡分布到网络中,达到流量负载均衡的目标。而且采用集中控制思想,减少了分布式系统下复杂的协议交互,能够有效提升网络故障恢复能力,有效支持故障下的流量工程优化需要。第二,基于FR-TE系统结构提出了在线域内流量工程优化算法MCPF,并提出了一种基于链路负载的故障快速恢复方法。针对流量优化问题,本文基于FR-TE系统支持在线的流量优化,提出了一种在线的流量工程优化算法MCPF,该算法基于全局网络拓扑和网络中链路的状态,通过更新和维护网络链路状态达到使流量均衡分布到网络中的目标。本文基于链路负载提出一种快速故障恢复方法,该方法考虑链路负载和网络故障检测之间的关系,通过对低负载链路故障检测时间提前,高负载链路则对其重新优化并预计算链路故障后的处理结果,达到快速故障恢复的目标。第三,设计并实现了面向故障恢复的域内流量工程的原型系统。通过对quagga路由软件进行功能修改和扩展,实现了面向故障恢复的域内流量工程的原型系统,并基于CORE平台搭建仿真测试环境对系统的功能和性能进行测试,充分验证了FR-TE系统的流量优化能力和对故障快速恢复能力。
[Abstract]:With the rapid development of the Internet, network applications emerge in endlessly, and more traffic and services are carried by the network, which poses a great challenge to the reliability of the network. However, network failures occur from time to time, which seriously affect the normal operation of the network and bring huge losses to users and network service providers. Researchers have done some research on network fault recovery, which is mainly based on connectivity to transfer traffic from fault path to recovery path, without considering the load balance after network recovery. It is easy to cause new network congestion; The research of traffic engineering can improve the utilization ratio of network resources, reduce the maximum utilization ratio of network links, and make the network traffic distribute evenly to the network. However, preplanning is generally used to optimize the flow over a period of time, and the real-time response to network failures is insufficient. With the development of network architecture, the concept of SDN gives us a new idea. Based on the centralized routing architecture of SDN, we propose an intra-domain traffic engineering system for fault recovery. The corresponding online optimization algorithm MCPF, is proposed to plan the flow request in the network according to the actual network operation, so that the network can achieve load balance, and when the network failure occurs, the path optimization can be carried out quickly for the fault. Quick recovery and load balancing after failure. The main work of this thesis is as follows: first, a fault recovery oriented intra-domain flow engineering system (FR-TE) architecture is proposed. Aiming at the optimal demand of flow engineering under the condition of fault, based on the idea of centralized control, the structure of flow engineering system in the domain for fault recovery is proposed. Through the real-time planning and deployment of the traffic requests on the network forwarding nodes, the system structure enables the network traffic to be evenly distributed to the network and achieves the goal of traffic load balancing. By using centralized control idea, the complex protocol interaction in distributed system can be reduced, and the network fault recovery ability can be effectively enhanced, and the flow engineering optimization needs can be effectively supported. Secondly, based on the structure of FR-TE system, an on-line flow engineering optimization algorithm, MCPF, is proposed, and a fast fault recovery method based on link load is proposed. Aiming at the traffic optimization problem, this paper proposes an online traffic engineering optimization algorithm, MCPF, which is based on the global network topology and the state of the link in the network, based on the FR-TE system supporting online traffic optimization. By updating and maintaining the network link state, the goal of evenly distributing traffic to the network is achieved. This paper proposes a fast fault recovery method based on link load, which considers the relationship between link load and network fault detection. The high-load link is reoptimized and the processing results after link failure are predicted to achieve the goal of fast fault recovery. Thirdly, a prototype system of intra-domain flow engineering for fault recovery is designed and implemented. By modifying and extending the function of quagga routing software, the prototype system of intra-domain traffic engineering for fault recovery is realized, and the function and performance of the system are tested by building a simulation test environment based on CORE platform. The flow optimization ability and fast recovery ability of FR-TE system are fully verified.
【学位授予单位】：国防科学技术大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP393.06

【相似文献】