發(fā)布時間：2019-06-14 13:14

【摘要】：隨著互聯網的高速發(fā)展,網絡已經深入到人們的各個方面,每條鏈路上承載流量也越來越重要,鏈路失效問題日漸突出。軟件定義網絡(Software Defined Network,SDN)是一種新型網絡架構,將數據平面和控制平面相分離,可以更靈活地部署網絡應用來解決鏈路失效問題。本文主要研究軟件定義網絡中鏈路失效問題。根據網絡中鏈路失效的規(guī)模,該問題可以分為單鏈路失效和多鏈路失效兩種情形,本文針對這兩種情形分別提出相應鏈路恢復算法。在單鏈路失效問題中,本文提出基于流聚合的本地快速重路由和基于擁塞避免的快速重路由兩種算法。基于流聚合的本地快速重路由算法利用軟件定義網絡中流表匹配的靈活性,如果一條鏈路發(fā)生失效,所有被中斷的業(yè)務會被聚合成一條“大”的業(yè)務,然后SDN控制器動態(tài)地部署一條本地恢復路徑代替失效鏈路。通過仿真結果可以得出,該算法可以減少SDN控制器更新流表項的數量,同時快速恢復被中斷業(yè)務。而基于擁塞避免的快速重路由是在前者的基礎上,考慮鏈路容量以及業(yè)務帶寬,將被中斷業(yè)務聚合為多個“中等”業(yè)務并部署在不同的本地恢復路徑上,從而有效避免業(yè)務恢復后造成鏈路擁塞的問題。但是該算法的失效恢復時延略微有所增加,所以在單鏈路失效發(fā)生時,需要根據網絡負載情況選擇相應的算法進行恢復。在多鏈路失效問題中,考慮到網絡拓撲改變較大而且網絡中大部分業(yè)務可能受到影響,逐一恢復每個被中斷業(yè)務的效率較低。本文提出基于生成樹的快速恢復算法,在鏈路失效后,以樹的形式連接所有有效交換機,在SDN控制器統(tǒng)一更新交換機流表后,所有受影響的業(yè)務會自動被恢復。利用樹結構的特殊性,在失效恢復過程中,可以減少控制器對流表操作,從而實現快速失效恢復。
[Abstract]:With the rapid development of the Internet, the network has penetrated into all aspects of people, and the traffic on each link is becoming more and more important, and the problem of link failure is becoming more and more prominent. Software defined Network (Software Defined Network,SDN) is a new network architecture, which separates the data plane from the control plane and can deploy network applications more flexibly to solve the link failure problem. In this paper, the problem of link failure in software-defined networks is studied. According to the scale of link failure in the network, the problem can be divided into single link failure and multi-link failure. In this paper, the corresponding link recovery algorithms are proposed for these two cases. In the problem of single link failure, this paper proposes two algorithms: local fast rerouting based on flow aggregation and fast rerouting based on congestion avoidance. The local fast rerouting algorithm based on flow aggregation makes use of the flexibility of software to define the matching of flow table in the network. If a link fails, all interrupted traffic will be aggregated into a "large" service, and then the SDN controller will dynamically deploy a local recovery path instead of the failed link. Through the simulation results, it can be concluded that the algorithm can reduce the number of update stream table items of SDN controller and restore the interrupted traffic quickly. The fast rerouting based on congestion avoidance is based on the former, considering link capacity and service bandwidth, the interrupted traffic is aggregated into multiple "medium" services and deployed on different local recovery paths, so as to effectively avoid the problem of link congestion caused by traffic recovery. However, the failure recovery delay of the algorithm increases slightly, so when the single link failure occurs, the corresponding algorithm needs to be selected according to the network load. In the problem of multi-link failure, considering that the network topology changes greatly and most of the traffic in the network may be affected, it is inefficient to restore each interrupted traffic one by one. In this paper, a fast recovery algorithm based on spanning tree is proposed. After link failure, all effective switches are connected in the form of tree. After the SDN controller updates the switch flow table uniformly, all affected services will be automatically restored. Taking advantage of the particularity of tree structure, the controller convective meter operation can be reduced in the process of failure recovery, so that the rapid failure recovery can be realized.
【學位授予單位】：電子科技大學
【學位級別】：碩士
【學位授予年份】：2017
【分類號】：TP393.02

【參考文獻】

相關期刊論文 前3條

1 俞慧春;;SDN技術的發(fā)展和應用淺析[J];中國新通信;2014年16期

2 韋世紅;盧威;;基于OpenFlow的未來互聯網試驗網[J];計算機系統(tǒng)應用;2013年05期

3 王淑玲;李濟漢;張云勇;房秉毅;;SDN架構及安全性研究[J];電信科學;2013年03期

相關碩士學位論文 前1條

1 彭澤鑫;SDN網絡抗毀問題研究[D];電子科技大學;2016年

，

本文編號：2499413