OpenFlow網(wǎng)絡中服務器負載均衡的研究
發(fā)布時間：2019-04-17 13:23
【摘要】：目前許多互聯(lián)網(wǎng)企業(yè)都需要為大量的用戶同時提供服務,這些企業(yè)通常需要建立一個復雜的網(wǎng)絡,并使用多臺服務器對外提供服務,為了合理的利用資源,降低成本,提供高質(zhì)量可靠的服務,企業(yè)內(nèi)部通常會使用負載均衡技術(shù),由于傳統(tǒng)互聯(lián)網(wǎng)自身的缺陷,現(xiàn)有的負載均衡價格昂貴,也缺乏足夠的靈活性。SDN的提出為許多現(xiàn)有的網(wǎng)絡問題提供了新的解決方案,本文利用SDN的一些特性,在OpenFlow網(wǎng)絡中設(shè)計并實現(xiàn)了服務器負載均衡。主要完成了以下工作： 1、目前OpenFlow交換機價格昂貴,基于NetFPGA的OpenFlow交換機成為實驗人員的首選,本文對基于NetFPGA的OpenFlow交換機進行了一系列的測試,為OpenFlow網(wǎng)絡的研究做了很好的準備工作。 2、在OpenFlow網(wǎng)絡中控制器通常由軟件實現(xiàn),對軟件來講性能往往是人們最關(guān)心的問題之一,本文測試了Floodlight控制器下發(fā)流表的性能。 3、設(shè)計并實現(xiàn)了一個基于服務器運行狀態(tài)的負載均衡算法,該算法根據(jù)服務器CPU和內(nèi)存的使用率計算服務器的綜合負載,當客戶端請求到達時,控制器選擇綜合負載最小的服務器為用戶提供服務。 4、設(shè)計了一種服務器負載均衡方案,該將網(wǎng)絡的管理和服務器的狀態(tài)監(jiān)控結(jié)合起來,OpenFlow交換機通過對數(shù)據(jù)包的修改把請求轉(zhuǎn)發(fā)給綜合負載最小的服務器,這樣可以有效的利用每一臺服務器的計算能力。
[Abstract]:At present, many Internet enterprises need to provide services to a large number of users at the same time. These enterprises usually need to establish a complex network and use multiple servers to provide services. In order to make rational use of resources and reduce costs, To provide high-quality and reliable services, load balancing technology is often used within the enterprise, and existing load balancing is expensive due to the shortcomings of the traditional Internet itself. The proposed SDN provides a new solution for many existing network problems. This paper designs and implements server load balancing in OpenFlow network by utilizing some characteristics of SDN. The main work is as follows: 1. At present, OpenFlow switches are expensive, NetFPGA-based OpenFlow switches have become the first choice for experimenters. In this paper, a series of tests on NetFPGA-based OpenFlow switches have been carried out. Good preparation has been made for the research of OpenFlow network. 2. In the OpenFlow network, the controller is usually implemented by software, and the performance of the controller is one of the most concerned problems for the software. This paper tests the performance of the flow table under the Floodlight controller. 3, a load balancing algorithm based on server running state is designed and implemented. The algorithm calculates the overall load of the server according to the usage of server CPU and memory. When the client requests arrive, the load of the server is calculated. The controller selects the server with the least combined load to serve the user. 4. A server load balancing scheme is designed, which combines the network management with server status monitoring. The OpenFlow switch forwards the request to the server with the least combined load by modifying the packet. This can effectively utilize the computing power of each server.
【學位授予單位】：蘭州大學
【學位級別】：碩士
【學位授予年份】：2014
【分類號】：TP393.05
本文編號：2459484
[Abstract]:At present, many Internet enterprises need to provide services to a large number of users at the same time. These enterprises usually need to establish a complex network and use multiple servers to provide services. In order to make rational use of resources and reduce costs, To provide high-quality and reliable services, load balancing technology is often used within the enterprise, and existing load balancing is expensive due to the shortcomings of the traditional Internet itself. The proposed SDN provides a new solution for many existing network problems. This paper designs and implements server load balancing in OpenFlow network by utilizing some characteristics of SDN. The main work is as follows: 1. At present, OpenFlow switches are expensive, NetFPGA-based OpenFlow switches have become the first choice for experimenters. In this paper, a series of tests on NetFPGA-based OpenFlow switches have been carried out. Good preparation has been made for the research of OpenFlow network. 2. In the OpenFlow network, the controller is usually implemented by software, and the performance of the controller is one of the most concerned problems for the software. This paper tests the performance of the flow table under the Floodlight controller. 3, a load balancing algorithm based on server running state is designed and implemented. The algorithm calculates the overall load of the server according to the usage of server CPU and memory. When the client requests arrive, the load of the server is calculated. The controller selects the server with the least combined load to serve the user. 4. A server load balancing scheme is designed, which combines the network management with server status monitoring. The OpenFlow switch forwards the request to the server with the least combined load by modifying the packet. This can effectively utilize the computing power of each server.
【學位授予單位】：蘭州大學
【學位級別】：碩士
【學位授予年份】：2014
【分類號】：TP393.05
【參考文獻】
相關(guān)期刊論文 前4條
1 余侃;;云計算時代的數(shù)據(jù)中心建設(shè)與發(fā)展[J];信息通信;2011年06期
2 陳娟,崔燦,凌云,陳綺;SNMP擴展代理的實現(xiàn)[J];計算機工程;2005年13期
3 左青云;陳鳴;趙廣松;邢長友;張國敏;蔣培成;;基于OpenFlow的SDN技術(shù)研究[J];軟件學報;2013年05期
4 薛軍,李增智,王云嵐;負載均衡技術(shù)的發(fā)展[J];小型微型計算機系統(tǒng);2003年12期
，本文編號：2459484
本文鏈接：http://lk138.cn/guanlilunwen/ydhl/2459484.html
最近更新
教材專著
