臺灣博碩士論文加值系統

現今的基礎網路架構與運行方式因需求的上升已漸漸不敷使用，因此開始有新的網路架構被提出，希望能符合未來網路的需求。目前最具代表性的未來網路架構是軟體定義是網路(Software Defined Network, SDN)。而且目前已有許多的研究人員從事SDN相關的研究。但是在單一實驗室中，想進行大規模的網路實驗是件困難的事情，而網路實驗平台則可以解決這個問題。於是本論文設計並實作一個SDN網路實驗平台。SDN網路實驗平台提供虛擬的OpenFlow Switch與OpenFlow Controller供研究人員進行OpenFlow相關之實驗。
為提升網路實驗的方便性，於是SDN網路實驗平台加入了網路功能虛擬化(Network Functions Virtualization，NFV)的概念。網路功能虛擬化可以將防火牆(Firewall)、Network address translation、深度封包檢測(Deep Packet Inspection)、閘道器(Gateway)等網路功能利用虛擬化技術變成虛擬網路功能(Virtual Network Function)。將虛擬網路功能加入網路實驗平台後，研究人員可以直接在實驗拓撲中新增虛擬網路功能。簡化自行建立虛擬網路功能的複雜程序。但是這些網路服務可能會在SDN網路實驗平台中產生大量的網路流量，若又發生網路功能和客戶端被分隔在距離遙遠的兩個計算機中心內，則會增加服務的延遲、有限的網路頻寬被嚴重的消耗等問題，為改善這個問題，本論文提出Shortest Path Algorithm-based Sliced Network Weighted Graph機制，藉由這個機制來搬移虛擬網路功能到客戶端附近的資料中心內，來減少網路服務的延遲與降低網路頻寬的消耗。在實驗中，使用本機制可以將傳出延遲從原本的平均3.21ms下降到0.46ms，改善幅度為85.5%的成效。

The basic network devices are not enough to use, so there are some new network architectures are proposed. New architectures can support the demands of future networks. Software Defined Network (SDN) is the most popular future network architecture.more and more scholars dedicate to SDN researchs. But doing large-scale network experiment is difficult in single laboratory, and the network testbed can solve this problem. It designs and implements an SDN-based network testbed in this paper. SDN-based netwrok testbed provides virtual OpenFlow Switch and OpenFlow Controller. Scholars can use those resources to do OpenFlow experiments.
The SDN-baded Network Testbed has the concept of Network Functions Virtualization(NFV) to enhance the convenience of the Internet experiments. NFV use virtualization technologies to create Virtual Network Functions(VNF). Such as Firewall, Network address translation, Deep Packet Inspection, Gateway. In the SDN-based Network Testbed, scholars have the ability to deploy VNF in their experiment topologies.But VNF may create large network traffic between Data Centers. So this paper, provide Shortest Path Algorithm-based Sliced Network Weighted Graph(SPA-SNWG) mchanism to reduce network traffic and latency. SPA-SNWG reduce network traffic and latency by migration VNF close to clients. In the experimenet, using SPA-SNWG mechanism can reduce network latency from 3.21ms to 0.46ms, improving the margin of 85.5% of the results.

第1章 緒論 1
1.1 概要 1
1.2 研究動機 3
1.3 研究目的 4
1.4 章節架構 4
第2章 背景知識與相關研究 5
2.1 Software Defined Network 與 OpenFlow 5
2.2 FlowVisor 8
2.3 網路功能虛擬化 10
2.4 網路測試平台 11
2.5 最短路徑演算法 13
2.6 相關研究之比較 14
第3章 SDN網路測試平台 17
3.1 平台架構與設計 17
3.1.1. Integrated Management Center模組 20
3.1.1.1. RSpec Integrate and Store模組 20
3.1.1.2. Web GUI模組 21
3.1.1.3. Topology Scheduling模組 21
3.1.2. Resource Management模組 22
3.1.2.1. VM Management模組 23
3.1.2.2. VM deployment模組 24
3.1.2.3. NFV Management模組 24
3.1.2.4. Resource Monitor模組 25
3.1.2.5. Monitor Agent模組 26
3.1.3. Slice Management模組 28
3.2 系統運作流程與Shortest Path Algorithm-based Sliced Network Weight Graph機制設計 29
3.2.1. 系統定義與假設 29
3.2.2. 資料符號表 30
3.2.3. 網路實驗平台系統功能運作流程 32
3.2.4. 基於SPA-SNWG之降低流量系統運作流程 36
第4章 實驗與討論 42
4.1 系統實作 42
4.2 情境一:網路實驗平台之圖形化介面操作 49
4.2.1. 實驗一：虛擬網路拓撲產生器 49
4.2.2. 實驗二：資源監視 56
4.3 情境二：NFV虛擬機器搬移 58
4.3.1. 實驗三：驗證假設「內部網路與外部網路比較」 58
4.3.2. 實驗四：NFV虛擬機器搬移表現 60
4.3.3. 實驗五：NFV虛擬機器搬移測試 62
第5章 結論與未來研究方向 65
5.1 結論 65
5.2 未來研究 66
參考文獻 68

[1] D. Mcdysan, "Software defined networking opportunities for transport," Communications Magazine, IEEE , vol.51, no.3, pp.28-31, March.2013
[2] A. Lara, A. Kolasani, B. Ramamurthy, "Network Innovation using OpenFlow: A Survey," Communications Surveys &; Tutorials, IEEE , vol.16, no.1, pp.493,512, First Quarter 2014
[3] OpenFlow, http://archive.openflow.org/
[4] White Papers：Software-Defined Networking, https://www.opennetworking.org/sdn-resources/sdn-library/whitepapers
[5] PlanetLab, https://www.planet-lab.org/
[6] Emulab.Net, http://www.emulab.net/
[7] GENI, www.geni.net
[8] NetworkCore – GENI, http://groups.geni.net/geni/wiki/NetworkCore
[9] F. Yue, "Network Functions Virtualization—Everything Old Is New Again.," 2013
[10] OpenFlow Specifications, https://www.opennetworking.org/sdn-resources/onf-specifications/openflow
[11] OpenFlow Switch Specification Version 1.3.4(protocol version 0x04), https://www.opennetworking.org/sdn-resources/onf-specifications/openflow
[12] OpenFlow Switch Specification Version 1.1.0, https://www.opennetworking.org/sdn-resources/onf-specifications/openflow
[13] Resource Specification(RSpec), http://groups.geni.net/geni/wiki/GENIExperimenter/RSpecs
[14] Shortest Path Problem - wikipedia, http://en.wikipedia.org/wiki/Shortest_path_problem
[15] Graph Theory – Wikipedia, http://en.wikipedia.org/wiki/Graph_theory
[16] Dijkstra Algorithm – Wikipedia, http://en.wikipedia.org/wiki/Dijkstra%27s_algorithm
[17] Bellman-Ford Algorithm – wikipedia, http://en.wikipedia.org/wiki/Bellman%E2%80%93Ford_algorithm
[18] Floyd-warshall Algorithm – Wikipedia, http://en.wikipedia.org/wiki/Floyd%E2%80%93Warshall_algorithm
[19] Li-Der Chou, Yao-Tsung Yang, Yuan-Mao Hong, Jhih-Kai Hu, Bill Jean, "A Genetic-Based Load Balancing Algorithm in OpenFlow Network," Advanced Technologies, Embedded and Multimedia for Human-centric Computing, pp.411-417, Aug. 2013
[20] Shi-Jia Peng, Li-Ming Tseng, "Botnet Victim Detection and Notification based on Openflow Switch," Jul. 2010
[21] H.E. Egilmez, S. Civanlar, AM. Tekalp, "An Optimization Framework for QoS-Enabled Adaptive Video Streaming Over OpenFlow Networks," Multimedia, IEEE Transactions on, vol.15, no.3, pp.710-715, Apr.2013
[22] M. Jarschel, F. Lehrieder, Z. Magyari, R. Pries, “A Flexible OpenFlow-Controller Benchmark,” Proceedings of the European Workshop on Software Defined Networking, pp.48-53, Oct. 2012
[23] Open vSwitch, http://openvswitch.org/
[24] 黃柏勝, 林冠圻, “OpenFlow暨OpenStack之多重Controller環境監測系統”, Jul. 2014