基于Linux系統(tǒng)的流量控制研究
發(fā)布時間：2019-03-27 12:04
【摘要】：流量控制常用于計算機網(wǎng)絡中,對網(wǎng)絡中的安全訪問進行有效的控制,并保障重要業(yè)務的帶寬。企業(yè)用戶常使用多網(wǎng)卡提高服務器的吞吐量、避免單點故障,以達到性能與經(jīng)濟成本的平衡。使用令牌桶算法在多網(wǎng)卡環(huán)境下進行流量控制時存在業(yè)務分類的問題,目前,缺少在這種場景下根據(jù)業(yè)務優(yōu)先級進行剩余帶寬利用的流量控制算法。另外,用戶通過流量控制配置工具實現(xiàn)流量控制,Linux系統(tǒng)上的流量配置工具tc未兼顧功能與用戶體驗兩方面,應針對Linux系統(tǒng)設計具有友好用戶接口的帶寬使用方法。 本文首先圍繞Linux系統(tǒng)上的流量控制機制進行研究,分析了Linux內(nèi)核中流量控制的流程與方法,對流量控制常用的FIFO、RED、TBF、CBQ、HTB等排隊規(guī)則算法總結(jié)了原理及特點；對排隊規(guī)則算法在內(nèi)核中的實現(xiàn)方式進行了分析；對流量控制工具tc與流量過濾工具iptables進行了介紹。針對上文提到的問題,本文主要完成了以下工作： 基于多網(wǎng)卡流量控制的需求,在令牌桶算法的基礎上,設計了改進的共享令牌桶算法STBF,并在實際應用場景中進行了實驗對比。改進的共享令牌桶算法能夠在多網(wǎng)卡環(huán)境中對不同的網(wǎng)卡設置服務優(yōu)先等級,在多個網(wǎng)卡間按照優(yōu)先級的高低實現(xiàn)空閑令牌的共享。實驗結(jié)果表明,改進的共享令牌桶算法能夠在多網(wǎng)卡應用環(huán)境中有效達到流量控制的目的,同時實現(xiàn)了按業(yè)務優(yōu)先級分配空閑令牌與TBF算法、HTB算法相比提高了帶寬利用率。在僅有優(yōu)先級高的網(wǎng)卡處于流量繁忙狀態(tài)時,STBF算法的平均帶寬利用率相比TBF算法由73%提升到了97%；同等負載狀態(tài)下,STBF算法的平均發(fā)送時延要低于TBF算法與HTB算法。 針對Linux流量配置工具tc的不足,結(jié)合tc的帶寬管理功能與iptables的流量過濾功能,設計了一種用于Linux系統(tǒng)的帶寬管理方法tctables,并對該方法進行了實驗測試。該方法具有容易掌握的使用方法、良好的移植性以及與比tc豐富的過濾功能,支持普通用戶、網(wǎng)絡管理員、編程開發(fā)人員等各類用戶進行帶寬管理或者帶寬管理軟件的開發(fā)。帶寬管理方法tctables包括tctables語言和tctables解釋器兩部分,文章中分別闡述了兩部分的工作原理與實現(xiàn)細節(jié)。文章中對使用tctables進行帶寬管理進行了測試,結(jié)果表明,通過tctables解釋器能夠編譯tctables語言寫成的腳本,最終能夠?qū)崿F(xiàn)期望的帶寬管理效果。
[Abstract]:Flow control is often used in computer networks to effectively control the secure access in the network and to guarantee the bandwidth of important services. Enterprise users often use multiple network cards to improve server throughput and avoid a single point of failure in order to achieve a balance between performance and economic costs. There exists the problem of traffic classification when using token bucket algorithm to control traffic in multi-network card environment. At present, there is no traffic control algorithm to utilize residual bandwidth according to service priority in this scenario. In addition, the user realizes the flow control through the flow control configuration tool, and the flow configuration tool tc on the Linux system does not take into account both the function and the user experience, so the bandwidth usage method with friendly user interface should be designed for the Linux system. In this paper, the flow control mechanism on Linux system is studied, the flow control flow and method in Linux kernel are analyzed, and the principle and characteristics of the queuing rules algorithm such as FIFO,RED,TBF,CBQ,HTB, which is commonly used in flow control, are summarized. The implementation of queue rule algorithm in kernel is analyzed, and the flow control tool tc and flow filter tool iptables are introduced. In view of the problems mentioned above, the main work of this paper is as follows: based on the demand of multi-network card flow control and based on token bucket algorithm, an improved shared token bucket algorithm STBF, is designed. The experimental results are compared with each other in the practical application scenario. The improved shared token bucket algorithm can set the service priority for different network cards in the multi-network card environment, and realize the sharing of idle tokens among the multiple network cards according to the priority level. The experimental results show that the improved shared token bucket algorithm can effectively achieve the purpose of flow control in the multi-network card application environment. At the same time, the allocation of idle tokens according to the service priority is realized compared with the TBF algorithm. Compared with the HTB algorithm, the bandwidth utilization is improved. When only high priority network cards are in a busy state, the average bandwidth utilization of STBF algorithm is increased from 73% to 97% compared with TBF algorithm, and the average transmission delay of STBF algorithm is lower than that of TBF algorithm and HTB algorithm under the same load condition. Aiming at the deficiency of Linux traffic configuration tool tc, combined with the bandwidth management function of tc and the traffic filtering function of iptables, a bandwidth management method tctables, for Linux system is designed and tested experimentally. This method has many advantages, such as easy to master, good portability and rich filtering function compared with tc. It supports all kinds of users, such as ordinary users, network administrators, programming developers and so on, to carry out bandwidth management or bandwidth management software development. The bandwidth management method tctables consists of two parts: tctables language and tctables interpreter. In this paper, the working principle and implementation details of the two parts are described. In this paper, the bandwidth management using tctables is tested. The results show that the script written in tctables language can be compiled by tctables interpreter, and the desired bandwidth management effect can be achieved at last.
【學位授予單位】：中國科學技術(shù)大學
【學位級別】：碩士
【學位授予年份】：2014
【分類號】：TP316.81;TP393.08
本文編號：2448160
[Abstract]:Flow control is often used in computer networks to effectively control the secure access in the network and to guarantee the bandwidth of important services. Enterprise users often use multiple network cards to improve server throughput and avoid a single point of failure in order to achieve a balance between performance and economic costs. There exists the problem of traffic classification when using token bucket algorithm to control traffic in multi-network card environment. At present, there is no traffic control algorithm to utilize residual bandwidth according to service priority in this scenario. In addition, the user realizes the flow control through the flow control configuration tool, and the flow configuration tool tc on the Linux system does not take into account both the function and the user experience, so the bandwidth usage method with friendly user interface should be designed for the Linux system. In this paper, the flow control mechanism on Linux system is studied, the flow control flow and method in Linux kernel are analyzed, and the principle and characteristics of the queuing rules algorithm such as FIFO,RED,TBF,CBQ,HTB, which is commonly used in flow control, are summarized. The implementation of queue rule algorithm in kernel is analyzed, and the flow control tool tc and flow filter tool iptables are introduced. In view of the problems mentioned above, the main work of this paper is as follows: based on the demand of multi-network card flow control and based on token bucket algorithm, an improved shared token bucket algorithm STBF, is designed. The experimental results are compared with each other in the practical application scenario. The improved shared token bucket algorithm can set the service priority for different network cards in the multi-network card environment, and realize the sharing of idle tokens among the multiple network cards according to the priority level. The experimental results show that the improved shared token bucket algorithm can effectively achieve the purpose of flow control in the multi-network card application environment. At the same time, the allocation of idle tokens according to the service priority is realized compared with the TBF algorithm. Compared with the HTB algorithm, the bandwidth utilization is improved. When only high priority network cards are in a busy state, the average bandwidth utilization of STBF algorithm is increased from 73% to 97% compared with TBF algorithm, and the average transmission delay of STBF algorithm is lower than that of TBF algorithm and HTB algorithm under the same load condition. Aiming at the deficiency of Linux traffic configuration tool tc, combined with the bandwidth management function of tc and the traffic filtering function of iptables, a bandwidth management method tctables, for Linux system is designed and tested experimentally. This method has many advantages, such as easy to master, good portability and rich filtering function compared with tc. It supports all kinds of users, such as ordinary users, network administrators, programming developers and so on, to carry out bandwidth management or bandwidth management software development. The bandwidth management method tctables consists of two parts: tctables language and tctables interpreter. In this paper, the working principle and implementation details of the two parts are described. In this paper, the bandwidth management using tctables is tested. The results show that the script written in tctables language can be compiled by tctables interpreter, and the desired bandwidth management effect can be achieved at last.
【學位授予單位】：中國科學技術(shù)大學
【學位級別】：碩士
【學位授予年份】：2014
【分類號】：TP316.81;TP393.08
【參考文獻】
相關(guān)期刊論文 前10條
1 李勇;周健;邵東軼;;一種基于Netfilter的BitTorrent流量控制方法[J];計算機安全;2008年04期
2 蔡文郁;金心宇;張昱;;基于Linux的網(wǎng)絡流量控制機制[J];江南大學學報;2006年04期
3 董劍安,王永剛,吳秋峰;iptables防火墻的研究與實現(xiàn)[J];計算機工程與應用;2003年17期
4 祝琳,王光彩,顧君忠;Linux網(wǎng)絡系統(tǒng)對QoS的支持[J];計算機工程;2002年02期
5 魏元雷;戚玉箐;胡西虹;;局域網(wǎng)系統(tǒng)基本性能指標的測試方法[J];計算機工程;2008年06期
6 湯庸,楊學良,區(qū)海翔,傅秀芬,李松;基于IP網(wǎng)絡的自適應QoS管理方案研究[J];計算機學報;2001年01期
7 樂艷輝;李之棠;柳斌;;基于Netfilter的P2P流量測量系統(tǒng)[J];計算機應用研究;2008年04期
8 孟曉景,井艷芳,張瑜;Linux內(nèi)核Netfilter防火墻原理與設計[J];山東科技大學學報(自然科學版);2004年02期
9 楊剛;陳蜀宇;;Linux中基于Netfilter/Iptables的防火墻研究[J];計算機工程與設計;2007年17期
10 何寶宏;IP網(wǎng)絡的服務質(zhì)量講座 第4講 IP網(wǎng)絡流量與擁塞控制技術(shù)[J];中國數(shù)據(jù)通信;2003年05期
相關(guān)博士學位論文 前1條
1 曾菊玲;寬帶無線接入網(wǎng)動態(tài)QoS機制研究[D];北京郵電大學;2010年
，本文編號：2448160
本文鏈接：http://sikaile.net/guanlilunwen/ydhl/2448160.html
最近更新
教材專著
