臺灣博碩士論文加值系統

雲端運算、大數據、物聯網、IT產業及5 G產業的趨起，使得傳統分散式的網路架構及提供雲端服務的供應商面臨諸多挑戰與瓶頸。例如，物聯網快速的成長使得網路的管理與監控成為網管人一項艱鉅的任務。因此，雲端資料中心的建置與控管，網路資源的整合、網路流量的優化以及如何提升網路應用服務的品質使其能達成最佳服務層級協議(Service Level Agreement, SLA)之需求已是各雲端服務供應商所積極要解決的問題。此外網路虛擬化技術在傳統網路架構中亦有諸多的限制。近年來，新一代網路架構軟體定義網路(Software-Defined Network, SDN) 的驅起，使得上述的問題相對的減少。SDN採用集中式的控管架構，具有動態可程式化及網路虛擬化功能等技術，且具備有橫向擴展(Scale-Out)與大量部署的能力，可為雲端服務業者降低管理部署人力與營運成本。因此，在雲端數據中心，傳統分散式的網路架構已逐漸被軟體定義網路(SDN)所取代。有見於此，本研究致力於 SDN 雲端運算中心優化機制之研究。論文主體分成2個部份，分別論述如下：
由於雲端數據中心可能位於許多地區，其內部的網路環境可能與傳統雲端數據中心有所不同，且虛擬機(Virtual Machine, VM)如何部署也將影響服務性能。本研究第1部份針對Eucalyptus雲端計算平台，提出了一種網路調度機制，用於了解用戶和相關VM之間的網路運作狀態，以改善雲端服務品質。本研究中，OpenFlow交換器被建置在雲端平台上，用以管理來自用戶和虛擬機的預期流。研究機制在相同子網域和不同子網域上設置了節點控制器。對於每個實驗，測試了兩種情景。一是節點控制器在正常網路條件下進行，另一個是在節點控制器處於有瓶頸事件下進行。研究結果顯示，在正常網路的同一子網中，本研究所提出的機制與Round Robin和Greedy相比，本研究的傳輸時間分別提高了2.78％和68.76％。在考慮有瓶頸事件時，所提出的機制可以將傳輸時間分別提高32.40％和60.25％。在正常網路情景的不同子網中，所提出的機制與Round Robin和Greedy機制相比，傳輸時間提高了21.71％和52.94％，而在不同子網路且有瓶頸事件存在下，所提出的機制之傳輸時間提高了23.52％和50.80％。
隨著雲端服務，IoT，Big Data和5G等數據處理對網路需服務的日益增長的求，軟件定義網路(SDN)已是目前大型企業(如Google，NEC)建置雲端服務中心不可或缺的工具。 SDN 提供一個新的網路介面具有俯視網路路由演算路徑、網路吞吐量及網路連通性的全局視圖。因此，本研究第2個部份旨提供了一種新穎的SDN動態路由演算法，稱為階層陣列路由演算法(HARP)。 HARP路由演算法在SDN-LAN中採用分層和陣列多路徑分配，可以有效降低封包傳送的等待間間。測試平台使用OpenDayLight作為雲端平台中OpenFlow控制器，並運用Mininet模擬器，OpenvSwitch，D-ITG和Wireshark構建完整雲端平台。模擬結果與OSPF和Hertiana所提出的基於流的路由(SBR)的結果進行比較。比較結果顯示，HARP具有最短的等待時間，而HARP的平均Overhead size比SBR小7％左右。然而，OSPF的Overhead size大小比HARP和SBR更好，因為更多的交換機有可能增加Overhead size。但整體實驗結果顯示，HARP演算法仍然優於SBR演算法。

The rise of Cloud Computing, Big Data, Internet of Things (IoT), Information Technology (IT) and 5G, presents the traditional decentralized network structure and cloud services providers with many challenges and bottlenecks. For instance, the rapid growth of Internet of Things makes network management and monitoring difficult. Therefore, to achieve Service Level Agreement (SLA) requirements, cloud service providers need to establish and control cloud data centers, integrate network resources, network traffic optimization and improve the quality of network application services. Additionally, network virtualization technology in the traditional network architecture also has many restrictions, some of which are alleviated by the new Software-Defined Network (SDN) architecture. An SDN adopts a centralize control architecture with technologies such as dynamic programmable and network virtualization capabilities, and can scale-out and deploy large volumes, thus reducing management, manpower and operation costs for cloud service providers. Therefore, SDN has gradually replaced the traditional decentralized network architecture in cloud data centers. This study explores the optimization mechanism of SDN cloud computing centers. The main body of this study is divided into two parts, which are discussed as follows.
As a cloud data center may be located over many regions, and may have a non-standard network environment, the deployment of Virtual Machines (VMs) influence service performance. The first part of this study presents a network-constrained scheduling mechanism that works on the current network status between users and associated VMs to improve cloud services, and applies this mechanism to the Eucalyptus cloud computing platform. An OpenFlow switch is implemented on a cloud controller to manage the intended flow from users and VMs. The mechanism is tested in two configurations, with the node controller on either the same subnet or a different subnet for the VM. Each configuration is tested in two scenarios, one conducted under normal network conditions, and the other under a bottleneck event at a cloud controller. Experimental results indicate that with the controller in the same subnet as the VMs in the normal network, the proposed mechanism improves transmission time by 2.78% and 68.76% over the Round Robin and Greedy mechanisms, respectively. The proposed mechanism improves transmission time in bottleneck events by 32.40% and 60.25% compared to the Round Robin and Greedy mechanisms, respectively. With the controller on a different subnet, the proposed mechanism improves transmission time by 21.71% and 52.94% compared to that of the Round Robin and Greedy mechanisms, respectively, in the normal network scenario, and by 23.52% and 50.80%, respectively, in the bottleneck event.
The increasing demands of network services from data processing, such as cloud services, IoT, Big Data and 5G mean that a Software-Defined Network (SDN) is an indispensable tool for large enterprises (such as Google, NEC) to construct cloud services. An SDN provides a novel web interface with a global view of overhead network routing algorithms, network throughput and network connectivity. Therefore, the second part of this study presents a novel SDN dynamic routing algorithm, called Hierarchical Array Routing Path algorithm (HARP). HARP applies hierarchical and array multi-path allocation in SDN-LAN, thus shortening the waiting time between packet transmission and the overhead size. A testbed is implemented using OpenDayLight as an OpenFlow Controller, and is constructed with the Mininet emulator, OpenvSwitch, D-ITG and Wireshark. Simulation results are compared with those of OSPF and Hertiana’s proposed Flow-Based Routing (SBR) algorithm. Comparison results demonstrate that HARP has the shortest waiting time, and also has an average overhead size about 7% smaller than SBR. However, OSPF has better overhead size reduction than HARP and SBR, because more switches means larger overhead. However, HARP still performs better overall than SBR.

摘 要 I
Abstract III
Contents VII
List of Figures IX
List of Tables XI
Chapter 1 Introduction 1
1.1 Motivation 1
1.2 Contribution 5
1.3 Organization of This Thesis 6
Chapter 2 Background Knowledge 7
2.1 Cloud Computing 7
2.2 Open-Source Cloud Computing Platforms 9
2.2.1 Eucalyptus Cloud Platforms 9
2.2.2 OpenStack Cloud Platforms 11
2.2.3 Cloud Virtualization 13
2.3 Main Architecture of SDN 14
2.3.1 OpenFlow Architectur 15
2.3.2 OpenFlow Controller 17
2.3.3 OpenFlow Protocol 18
2.3.4 OpenFlow Switch 19
2.3.5 OpenFlow Channel 19
2.3.6 Flow Table 20
2.3.7 Group Table 21
2.3.8 Multiple Flow Table 21
Chapter 3 Proposed System Architecture and System Algorithm 23
3.1 Proposed Network-Constrained VM Deployment (NCVMD) Mechanism 23
3.1.1 VMs Scheduling Algorithm 24
3.1.2 NC Selection Mechanism 28
3.2 Performance Analysis of Hierarchical Array Routing Path on SDN 29
3.2.1 System Overview 29
3.2.2 Algorithm Operation 30
3.2.3 Algorithm Proof 32
3.2.4 Demonstration that HARP has a Shorter Waiting Time than OSPF 34
3.2.5 Extending HARP to K Routing Paths 38
3.2.6 Computer Simulation to Estimate The Real Network Packet Transmission 40
Chapter 4 System Performance Analysis 45
4.1 System Implementation of NCVMD 45
4.2 Performance Analysis of NCVMD 48
4.3 System Implementation of HARP 60
4.4 Simulation Result Analysis 62
Chapter 5 Conclusion and Future Work 65
5.1 Conclusion 65
5.2 Future Work 66
Reference 68
Appendix A: System Set Up 73
A-1: Installation of Ubuntu 14.04 and Java JDK 73
A-2: Installation of OpenFlow Switch 76
A-3: Installation of Wireshark 78
A-4: Installation of Mininet 79
A-5: Installation of Mininet 80
A-6: Mininet Testing 84
A-7: Opendaylight Testing 85
A-8: Eucalyptus Set Up 86

[1] C. P. Tang, P. P. C. Lee,, and T. Y. Wong, “Tunable Version Control System for Virtual Machines in an Open-Source Cloud,” Proceedings of the IEEE Transactions On Services Computing, pp.155-168, Feb 2015.
[2] K. Bakshi, “Considerations for Cloud Data Center: Framework, architecture and adoption,” Proceedings of the IEEE Aerospace Conference, pp.1-7, Mar 2011.
[3] S. M. Deshpande and B. Ainapure, “An Intelligent Virtual Machine Monitoring System Using KVM for Reliable And Secure Environment in Cloud,” Proceedings of IEEE International Conference on Advances in Electronics, Communication and Computer Technology, pp.314-319, Dec 2016.
[4] Netadmin, Available: http://www.netadmin.com.tw
[5] A. S. Ameer Mosa, A. S. Ali, J. Xue, and T. Scott, “Routing algorithm optimization for software defined network WAN,” Proceedings of Al-Sadeq International Conference on Multidisciplinary in IT and Communication Science and Applications (AIC-MITCSA), pp.1-6. May 2016.
[6] V. Adrichem, N. L. M., C. Doerr, and F. A. Kuipers, “OpenNetMon: Network monitoring in OpenFlow Software-Defined Networks,” Proceedings of the IEEE Network Operations and Management Symposium (NOMS), pp.1–8, May 2014.
[7] K. G. Yalda, D. J. Hamad, and I. T. Okumus, “Design and Implementation of an Intra-domain routing module for an SDN controllerfor Traffic Engineering in SDN environment,” proceedings of 2015 International Conference on Advances in Software, Control and Mechanical Engineering (ICSCME-2015), pp.1-7, Sep 2015
[8] N. Mckeown, T. Anderson, H. Balakrishnan, G. Paruldar, L. Peterson, J. Rexford, S. Shenker, and J. Turner, “OpenFlow: Enabling Innovation in Campus Networks,” Proceedings of the ACM SIGCOMM Computer Communication, Vol.38, Apr 2008.
[9] J. Pang, G. Xu, and X. Fu, “SDN-Based Data Center Networking With Collaboration of Multipath TCP and Segment Routing,” Proceedings of IEEE Access, pp.9764-9773, May 2017.
[10] B. Lantz, B. Heller, and N. McKeown, “A network in a laptop:rapid prototyping for software-defined networks,” Proceedings of the Ninth ACM SIGCOMM Workshop on Hot Topics in Networks, pp.1–6, Oct 2010.
[11] S. Oda, and D. Nobayashi, Yutaka Fukuda and Takeshi Ikenaga, “Flow-based Routing Schemes for Minimizing Network Energy Consumption using OpenFlow,” Proceedings of the Fourth International Conference on Smart Grids, Green Communications and IT Energy-aware Technologies, pp.20 – 24, Apr 2014.
[12] S. N. Hertiana, Hendrawan, and A. Kurniawan, “Performance analysis of flow-based routing in software-defined networking,” Proceedings of 22nd Asia-Pacific Conference on Communications (APCC), pp.579-585, May 2016.
[13] M. Bredel, Z. Bozakov, A. Barczyk, and H. Newman, “Flow-Based Load Balancing in Multipathed Layer-2 Networks using OpenFlow and Multipath-TCP,” Proceedings of ACM HotSDN, pp.213-214, Aug 2014.
[14] H. E. Egilmez, S. T. Dane, K. T. Bagci, and A. M. Tekalp, “OpenQoS:An OpenFlow controller design for multimedia delivery with end-to-end Quality of Service over Software-Defined Networks,” Proceedings of Signal & Information Processing Association Annual Summit and Conference (APSIPA ASC), pp.1-8, Dec 2012.
[15] F. Qin, Z. Zhao and H. Zhang, “Optimizing Routing and Server Selection in Intelligent SDN-based CDN,” Proceedings of 8th International Conference on Wireless Communications & Signal Processing (WCSP), pp.1-5, Oct 2016.
[16] TechNews, Available: https://technews.tw/2016/10/20/gartner-2017.../
[17] Y. Ran, J. Yang, S. Zhang, and H. Xi, “Dynamic IaaS Computing Resource Provisioning Strategy with QoS Constraint,” Proceedings of IEEE Transactions on Services Computing, pp.190-202, Mar. 2017.
[18] E. Sousa, F. Lins, E. Tavares, P. Cunha, and Maciel, “A Modeling Approach for Cloud Infrastructure Planning Considering Dependability and Cost Requirements,” Proceedings of IEEE International Conference on Systems, Man, and Cybernetics, pp.2168-2216, Sep 2014.
[19] M. J. Sule, M. Li, and G. Taylor, “Trust Modeling in Cloud Computing,” Proceedings of IEEE Symposium on Service-Oriented System Engineering (SOSE), pp.60-65, March 2016.
[20] Wiki, Available: https://en.wikipedia.org/wiki/Infrastructure_as_a_service
[21] Eucalyptus, Available: http://www.eucalyptus.com
[22] Y. Li, J. Zhang, Q. Hu, and J. Pei, “Research and practice on the theory of private clouds migration,” Proceedings of IEEE 13th International Conference on Signal Processing (ICSP), pp.1813-1818, March 2017.
[23] Y. Wang, and X. Li “Analysis of the Performance of Hadoop Applications on Eucalyptus Cloud,” Proceedings of IEEE International Symposium on Security and Privacy in Social Networks and Big Data (SocialSec), pp.28-31, Jan 2016.
[24] V. Amiry, S.Z. Rad, M.K. Akbari and M.S. Javan, “Implementing Hadoop Platform on Eucalyptus Cloud Infrastructure,” Proceedings of the 3PGCIC Seventh International Conference on P2P, Parallel, Grid, Cloud and Internet Computing, pp.74-78, Nov 2012.
[25] Z. Gao, and X. Li “Service QoS monitoring and analyzing based on Eucalyptus,” Proceedings of IEEE 4th International Conference on Computer Science and Network Technology (ICCSNT), pp.146-149, Jun 2016.
[26] G. R. Gangadharan,“Open Source Solutions for Cloud Computing,” Proceedings of the Computer, Vol.50, Jan 2017.
[27] J. L. Chen, W. H. Hsieh, W. K. Hsieh, and F. Y. Chou “Network-Constrained VMs Deployment Mechanism in Open Cloud Platform," Jokull Journal, pp. 241-259,Vol.65, Issue 5, May 2015
[28] OpenStack, Available: https://zh.wikipedia.org/wiki/OpenStack
[29] Z. W. Lei, D. M. Zhao, Y. T. Shi, and D. H. Sun “Research on the architecture and application of industrial cloud experimental platform based on OpenStack,” Proceedings of IEEE 29th Chinese Control and Decision Conference (CCDC), pp.7401-7406, Jul 2017.
[30] W. Zhan, L. Ruan; X. Yue, Z. Xu, and L. Xiao “A Secure and VM-supervising VDI System Based on OpenStack,” Proceedings of IEEE 7th International Conference on Cloud Computing and Big Data (CCBD), pp.187-192, Jul 2017.
[31] Z. Xu, L. Xiao; W. Zhan, X. Yue, L. Ruan, and R. Liu “A Secure and VM-supervising VDI System Based on OpenStack,” Proceedings of IEEE 7th International Conference on Cloud Computing and Big Data (CCBD), pp.58-63, Jul 2017.
[32] M. Stein, M. Scharf, and V. Hilt “SDN policy-driven service chain placement in OpenStack,” Proceedings of IFIP/IEEE Symposium on Integrated Network and Service Management (IM), pp.760-765, Jul 2017.
[33] I. Melinte, A. Bujor,, R. Dobre, and Herisanu, “Running KVM Virtual Machines in Batch Systems,” Proceedings of the 19th International Conference on Control Systems and Computer Science, pp.106-112, Jul 2013
[34] A.C. Bujor, and R. Dobre “KVM IO Profiling,” Proceedings of the 2013 RoEduNet International Conference, pp.26-28, Jan 2013
[35] B. Xu, Z. Peng, W. Ke, and M. Zhong “Deployment method of VM cluster based on graph theory for cloud resource management,” IET Journals & Magazines, pp. 622-627, Vol.11, Issue 5, May 2017
[36] Z. Zhou and L. Mu “An optimized virtual machine templates management mechanism for an Cloud system based on K-medoids Clustering,” Proceedings of the IEEE 35th Chinese Control Conference (CCC), pp.5243-5248, Aug 2016
[37] W. K. Hsieh, W. H. Hsieh, J. L. Chen F. Y. Chou, and Y. S. Lee, “Load Balancing Virtual Machines Deployment Mechanism In SDN Open Cloud Platform,” Proceedings of IEEE International Conference on Advanced Communication Technology (ICACT ), pp.3290-334, Aug, 2015.
[38] W. K. Hsieh, W. H. Hsieh, J. L. Chen, P. J. Yang “CssQoS: A Load Balancing Mechanism for Cloud Serving Systems,” Proceedings of the International Conference on Information Technology and Applications (ITA2014), pp.269-275, Jan, 2015.
[39] A.M. Lonea, D.E. Popescu, and Prostean, “A Survey of Management Interfaces for Eucalyptus Cloud,” Proceedings of the 7th IEEE International Symposium on Applied Computational Intelligence and Informatics, pp.261-266, Jul 2012
[40] Wiki, Available: https://en.wikipedia.org/wiki/Software_as_a_service
[41] Lara, A., Kolasani, A. and Ramamurthy, B. “Network Innovation using OpenFlow: A Survey,” proceedings of IEEE Communications Surveys & Tutorials, pp.493-512, Aug 2014
[42] Opennetworking.Org.2013, Available: https://www.opennetworking.org/sdn-resources/onf-specifications
[43] Opennetworking.Org.2014, Available: https://www.opennetworking.org/sdn-resources/onf-specifications
[44] OpenFlow, Available: http://archive.openflow.org/
[45] L. Han, Z. Li, W. Liu, K. Dai, and W. Qu, “Minimum Control Latency of SDN Controller Placement,” proceedings of IEEE Trustcom/BigDataSE/ISPA, pp.2175-2180, Feb 2017.
[46] Open Networking Foundation, Available: https://www.opennetworking.org/
[47] V. B. Wijekoon, T. M. Dananjaya, P. H. Kariyawasam, S. Iddamalgoda, and A. Pasqual, “High performance flow matching architecture for OpenFlow data plane,” proceedings of IEEE Conference on Network Function Virtualization and Software Defined Networks (NFV-SDN), pp.186-191, May 2017.
[48] OpenDaylight, Available: https://www.opendaylight.org/
[49] S. Koh, J. Kim, and S. Lee, “A proposal of OpenFlow controller to improve transfer rate in mesh network,” proceedings of International Conference on Information Networking (ICOIN), pp.509-511, Apr 2017.
[50] B. Ho, C. Pham-Quoc, T. N. Thinh, and N. Thoai, “A Secured OpenFlow-Based Switch Architecture,” proceedings of International Conference on Advanced Computing and Applications (ACOMP), pp.83-89, Jan 2017.
[51] Opennetworking.Org. 2014, Available: https://www.opennetworking.org/sdn-resources/onf-specifications
[52] Wiki, Available: https://en.wikipedia.org/wiki/OpenFlow
[53] Sofia Naning Hertiana, Hendrawan, and Adit Kurniawan, “Performance analysis of flow-based routing in software-defined networking,” proceedings of 22nd Asia-Pacific Conference on Communications (APCC), pp.579-585, Oct 2016
[54] M. Bredel, Z. Bozakov, A. Barczyk, and H. Newman, “Flow-Based Load Balancing in Multipathed Layer-2 Networks using OpenFlow and Multipath-TCP,” proceedings of ACM HotSDN, pp.213-214, Aug 2014.
[55] P. Habibi, M. Mokhtari, and M. Sabaei, “E-balance: An energy aware load balancer based on distributed OpenFlow controllers,” proceedings of 24th Iranian Conference on Electrical Engineering (ICEE), pp.1740-1745, Oct 2016
[56] S. Zeng, P. Zheng, and Y. Zhang, “Design of Test Case for OpenFlow Protocol Conformance Test Based on OFTest,” proceedings of International Symposium on Computer, Consumer and Control (IS3C), pp.465-470, Aug 2016.