【摘要】：軟件定義網(wǎng)絡(luò)(Software Defined Networking, SDN)作為一種新型的網(wǎng)絡(luò)架構(gòu),通過網(wǎng)絡(luò)應(yīng)用層、控制層及數(shù)據(jù)層的虛擬化和分離,實(shí)現(xiàn)了整個(gè)網(wǎng)絡(luò)的集中、動(dòng)態(tài)和可編程控制,使得整個(gè)網(wǎng)絡(luò)資源的管理、設(shè)計(jì)及使用變得更加靈活,也有利于網(wǎng)絡(luò)技術(shù)的革新與發(fā)展。 SDN集中化的網(wǎng)絡(luò)控制平臺(tái)給網(wǎng)絡(luò)帶來(lái)了諸如可編程性、易于管理、高適應(yīng)性、供應(yīng)商獨(dú)立等優(yōu)點(diǎn),但同時(shí)也產(chǎn)生了可擴(kuò)展性和可靠性方面的問題,因此,出現(xiàn)了分布式控制平臺(tái)的研究。本文提出了網(wǎng)絡(luò)設(shè)備受多個(gè)控制器控制時(shí)的松弛管理思想：當(dāng)網(wǎng)絡(luò)設(shè)備不參與部署某個(gè)控制器的虛擬網(wǎng)達(dá)到一定數(shù)量時(shí),控制器與網(wǎng)絡(luò)設(shè)備暫停通信,一段時(shí)間后,兩者再次通信。這提高了控制器與網(wǎng)絡(luò)設(shè)備之間的通信效率和控制器的管理效率�；谏鲜鏊枷�,本文首先設(shè)計(jì)了控制器松弛管理方案和算法；其次,通過建立網(wǎng)絡(luò)設(shè)備參與率、虛擬網(wǎng)絡(luò)部署成功率、網(wǎng)絡(luò)設(shè)備與控制器之間通信凈收益數(shù)學(xué)模型,對(duì)控制器松弛管理算法優(yōu)勢(shì)進(jìn)行定量分析。 虛擬網(wǎng)絡(luò)映射問題所涉及的是如何使有限的底層物理網(wǎng)絡(luò)資源支撐盡可能多的虛擬網(wǎng)絡(luò)。本文在SDN虛擬網(wǎng)絡(luò)映射中共進(jìn)行了兩個(gè)方面的研究,其一是,結(jié)合IETF (Internet Engineering Task Force,互聯(lián)網(wǎng)工程任務(wù)組)ForCES (Forwarding and Control Element Separation,轉(zhuǎn)發(fā)與控制分離)工作組的研究經(jīng)驗(yàn),在原有的虛擬網(wǎng)絡(luò)映射問題中引入了LFB (Logical Functional Block,邏輯功能塊)的概念,設(shè)計(jì)了功能塊共享方案,建立了基于功能塊共享的虛擬網(wǎng)絡(luò)映射算法,仿真結(jié)果顯示,本算法顯著降低了虛擬網(wǎng)絡(luò)映射過程中功能塊的租用費(fèi)用。其二是,首先,識(shí)別了優(yōu)先選擇剩余資源多或者少的物理節(jié)點(diǎn)這兩種映射方式,分別對(duì)其進(jìn)行了數(shù)學(xué)建模,實(shí)驗(yàn)結(jié)果顯示,優(yōu)先選擇剩余資源多的物理節(jié)點(diǎn)的映射的算法可以獲得較好的性能；其次,基于排隊(duì)論中損失制系統(tǒng)理論,對(duì)節(jié)點(diǎn)映射進(jìn)行數(shù)學(xué)建模,實(shí)驗(yàn)結(jié)果顯示,模型能夠較好地預(yù)測(cè)虛擬網(wǎng)絡(luò)映射性能。
[Abstract]:As a new network architecture, the software defined network (Software Defined Networking, SDN) realizes the centralized, dynamic and programmable control of the whole network through the virtualization and separation of network application layer, control layer and data layer, and makes the whole network resource management. Design and use become more flexible, also conducive to the innovation and development of network technology. SDN centralized network control platform to bring to the network such as programmability, easy to manage, high adaptability, independent suppliers and so on. But at the same time, there are some problems in scalability and reliability, so the research of distributed control platform appears. In this paper, the relaxation management idea of network equipment controlled by multiple controllers is proposed: when the network equipment does not participate in the deployment of a certain number of virtual networks, the controller and the network equipment stop communication, after a period of time, The two communicate again. This improves the communication efficiency between controller and network equipment and the management efficiency of controller. Based on the above ideas, this paper first designs the controller relaxation management scheme and algorithm; secondly, through the establishment of network equipment participation rate, virtual network deployment success rate, network equipment and controller communication between the net income mathematical model, The advantage of controller relaxation management algorithm is analyzed quantitatively. The problem of virtual network mapping is how to support as many virtual networks as possible by limited physical network resources. In this paper, two aspects of SDN virtual network mapping are studied. One is to combine the research experience of the IETF (Internet Engineering Task Force, Internet Engineering Task Force () ForCES (Forwarding and Control Element Separation, forwarding and Control Separation) working Group. The concept of LFB (Logical Functional Block, logical function block is introduced into the original virtual network mapping problem, and the scheme of function block sharing is designed, and the virtual network mapping algorithm based on functional block sharing is established. The simulation results show that, This algorithm can significantly reduce the rental cost of the function blocks in the virtual network mapping process. The other is that, firstly, we identify the two mapping methods of selecting more or less physical nodes with more or less remaining resources, and model them separately, and the experimental results show that, The algorithm of preferentially selecting the mapping of physical nodes with more remaining resources can obtain better performance. Secondly, based on the theory of loss system in queuing theory, the mapping of nodes is modeled mathematically, and the experimental results show that, The model can well predict the performance of virtual network mapping.
【學(xué)位授予單位】：浙江工商大學(xué)
【學(xué)位級(jí)別】：碩士
【學(xué)位授予年份】：2014
【分類號(hào)】：TP393.01

【參考文獻(xiàn)】

相關(guān)期刊論文 前10條

1 張燕平;李東亮;趙姝;鄭堯軍;;一種新的網(wǎng)絡(luò)拓?fù)浣７椒╗J];計(jì)算機(jī)工程與應(yīng)用;2010年22期

2 李仲學(xué),趙文廣,李翠平;離散事件仿真及其應(yīng)用的某些新進(jìn)展[J];計(jì)算機(jī)仿真;1999年04期

3 高明;王偉明;;ForCES體系結(jié)構(gòu)中LFB的研究與實(shí)現(xiàn)[J];計(jì)算機(jī)應(yīng)用研究;2006年03期

4 馮志偉;;齊普夫定律的來(lái)龍去脈[J];情報(bào)科學(xué);1983年02期

5 張順利;邱雪松;潘亞蓮;孟洛明;;網(wǎng)絡(luò)虛擬化環(huán)境下基于預(yù)測(cè)的資源重配置算法[J];通信學(xué)報(bào);2011年07期

6 羅娟;徐岳陽(yáng);李仁發(fā);;網(wǎng)絡(luò)虛擬化中動(dòng)態(tài)資源分配算法研究[J];通信學(xué)報(bào);2011年07期

7 張春環(huán);雷蕾;紀(jì)越峰;;一種新的網(wǎng)絡(luò)仿真拓?fù)渖善鱗J];系統(tǒng)仿真學(xué)報(bào);2006年11期

8 蔡宏堅(jiān);洪錦良;董黎剛;陳坷;王偉明;;ForCES路由器中邏輯功能塊應(yīng)用研究和實(shí)現(xiàn)[J];信息工程大學(xué)學(xué)報(bào);2009年01期

9 張潔,朱莉娟;貪心算法與動(dòng)態(tài)規(guī)劃的比較[J];新鄉(xiāng)師范高等�？茖W(xué)校學(xué)報(bào);2005年05期

10 付永紅;畢軍;吳建平;陳澤;王可;羅敏;;基于軟件定義網(wǎng)絡(luò)的一種多控制器休眠模型(英文)[J];中國(guó)通信;2014年03期

相關(guān)博士學(xué)位論文 前1條

1 卿蘇德;網(wǎng)絡(luò)虛擬化映射算法研究[D];北京郵電大學(xué);2013年

，

本文編號(hào)：2284132