隨著網路科技普及，跨國性的網路服務需求也隨之增加，為了提升服務品質，實現分散式伺服器架構及應用是一個可行的方案。另外近幾年面對海量的網路服務資源需求下，使得應用服務提供者採取伺服器叢集架構來達成負載平衡的目標，若結合上述兩種架構的優點，將會帶來用戶端與伺服器指派與封包路由的綜合問題。本研究針對分散式伺服器叢集架構中的用戶端與伺服器指派問題與封包路由問題定義出最佳化數學模型，並且證明出封包路由問題可獨立解決，接著提出修改式匈牙利演算法進行指派。為了實現本系統，我們提出在管理層實施修改式匈牙利演算法以及在控制層實施最短路徑封包路由方式的二層式控制架構並將此系統建構於軟體定義網路環境中。其中管理層獨立於控制層，我們提出採用網路功能虛擬化，將不同類型的分散式伺服器叢集管理實現在不同的虛擬網路功能上，以提升系統的實作彈性。此外本系統提供介面讓網路服務提供者做設定，根據設定內容，使系統可同時管理多個不同類型的分散式伺服器叢集。實驗結果顯示在二層式控制架構下，在控制層固定實施最短路徑封包路由方式，管理層中使用修改式匈牙利演算法做為用戶端與伺服器指派方式比起過去常見的負載平衡演算法在網路能量消耗及用戶端與伺服器之間的封包延遲的效能上皆來得好。

With the more popularity of Internet technology, the requirements of cross-border network services also increase. Applying the distributed server architecture is a viable option to improve Quality of Service. In addition, to satisfy the massive demand of network service resources in recent years, the application service provider adopts the server cluster architecture to achieve the goal of load balancing. Combining the advantages of the two architectures will introduce an integrated problem consisting of assignment problem and packet routing problem between clients and servers. In this thesis, we define the optimization model for the client and server assignment problem and packet routing problem in the distributed server cluster architecture, prove that the packet routing problem can be solved independently, and propose the revised Hungarian algorithm to solve the client and server assignment problem. Then, we design a two-layer control architecture to solve the optimization problem in the SDN environment. The two-layer control architecture is composed of the control plane and the management plane. The revised Hungarian algorithm is implemented in the management plane to solve the client and server assignment problem, and the shortest path packet routing is implemented in the control plane to solve the packet routing problem. Especially, we propose the use of Network Function Virtualization to manage different distributed server cluster services to enhance the scalability of the system. In addition, the management plane provides the interface for the network service provider to set the configuration of the distributed server cluster service, so that the proposed architecture can serve all requirements of the network service providers at the same time. The experimental results show that based on the shortest path packet routing in the control plane and various load balancing algorithms in the management plane, the performance of our revised Hungarian algorithm is better than that of other load balancing algorithms in terms of energy consumption and packet delay.

[1] S. Floyd and K. Fall, "Promoting the use of end-to-end congestion control in the Internet," IEEE/ACM Transactions on Networking, vo1.7, no.4, pp. 458-72, 1999.
[2] V. J. Ribeiro, R. H. Riedi, R. G. Baraniuk, J. Navratil and L. Cottrell, "pathChirp: Efficient Available Bandwidth Estimation for Network Paths," Proc. Passive and Active Measurements Workshop, pp. 1-11, Apr. 2003.
[3] W. Cheng, P. Shi and Z. Lei, "Network-assisted congestion control," Info-tech and Info-net Iternational Conferences, vol. 2, pp. 28-32, 2001.
[4] C. Sun, E. Stevens-Navarro, and V. W. S. Wong, "A constrained MDP based vertical handoff decision algorithm for 4G wireless networks," Proc. IEEE International Conference Communications, pp. 2169–2174, 2008.
[5] L. Wang and G. S. Kuo, "Mathematical modeling for network selection in heterogeneous wireless networks—A tutorial," IEEE Communications Surveys and Tutorials, vol. 15, no. 1, pp. 271–292, 2013.
[6] Z. Du, Q. Wu and P. Yang, "Learning with handoff cost constraint for network selection in heterogeneous wireless networks," Wiley's Journal of Wireless Communications and Mobile Computing (WCMC), vol. 16, pp. 441-458, 2016
[7] A. Keshavarz-Haddad, E. Aryafar, M. Wang and M. Chiang, "HetNets Selection by Clients: Convergence, Efficiency, and Practicality," IEEE/ACM Transactions on Networking, vol. 25, pp. 406-419, 2017
[8] M. Tarighi, S. A. Motamedi, S. Sharifian, "A new model for virtual machine migration in virtualized cluster server based on Fuzzy Decision Making," Journal of Telecommunications, vol. 1, pp. 40-51, 2010.
[9] D. M. Dias, W. Kish and R. Mukherjee, "A scalable and highly available web server," Compcon '96. 'Technologies for the Information Superhighway' Digest of Papers, pp. 1-8, 2002.
[10] V. Cardellini, M. Colajanni and P. S. Yu, "Dynamic load balancing on web-server systems," IEEE Internet Computing, pp. 28-39, 1999.
[11] A. F. R. Trajano and M. P. Fernandez, "Two-phase load balancing of In-Memory Key-Value Storages using Network Functions Virtualization (NFV)," Journal of Network and Computer Applications, vol. 69, pp. 1-13, 2016.
[12] A. F. R. Trajano and M. P. Fernandez, " Two-phase load balancing of In-Memory Key-Value Storages through NFV and SDN," 20th IEEE Symposium on Computers and Communication (ISCC), pp. 258-263, 2015.
[13] AFR Trajano and MP Fernandez, "uLoBal: Enabling In-Network Load Balancing for Arbitrary Internet Services on SDN," ICN, pp. 62-67, 2016.
[14] M. Aron, D. Sanders, P. Druschel and W. Zwaenepoel, "Scalable Content-aware Request Distribution in Cluster-based Network Servers," Proceedings of the 2000 USENIX Annual Technical Conference, pp. 1-15, 2000.
[15] A. V. Papadopoulosa, C. Klein, M. Maggio, J. D ürango, M. Dellkrantz, F. Hernández-Rodriguez, E. Elmroth, K.-E. Årzén, " Control-based load-balancing techniques: Analysis and performance evaluation via a randomized optimization approach," Control Engineering Practice, vol.52, pp. 24-34, 2016.
[16] S. M. Baker and B. Moon, "Distributed cooperative Web servers," Computer Networks, vol. 31, pp. 1215-1229, 1999.
[17] "SDN Architecture," Open Networking Foundation, 2016. Available on June 14, 2017: https://www.opennetworking.org/images/stories/downloads/sdn-resources/technical-reports/TR-521_SDN_Architecture_issue_1.1.pdf
[18] "SDN Architecture for Transport Networks," Open Networking Foundation, 2016. Available on June 14, 2017: https://www.opennetworking.org/images/stories/downloads/sdn-resources/technical-reports/SDN_Architecture_for_Transport_Networks_TR522.pdf
[19] "OpenFlow switch specification v1.3.0," Open Networking Foundation, 2012. Available on June 14, 2017: https://www.opennetworking.org/images/stories/d ownloads/sdn-resources/onf-specifications/openflow/openflow-spec-v1.3.0.pdf
[20] B. Han, V. Gopalakrishnan and L. Ji, "Network Function Virtualization : Challenges and Opportunities for Innovations," IEEE Communications Magazine, vol. 53, no. 2, pp. 90–97, 2015.
[21] J. Blendin, J. Rückert, N. Leymann, G. Schyguda and D. Hausheer, "Position Paper: Software-Defined Network Service Chaining," Third European Workshop Software Defined Networks (EWSDN), pp. 109–114, 2014.
[22] G. Lodi, F. Panzieri, D. Rossi, and E. Turrini, "SLA-Driven Clustering of QoS-Aware Application Servers," IEEE Transactions on Software Engineering, vol. 33, pp. 186-197, 2007.
[23] Z. Shang, W. Chen, Q. Ma and B. Wu, "Design and implementation of server cluster dynamic load balancing based on OpenFlow," International Joint Conference on Awareness Science and Technology and Ubi-Media Computing, pp. 691-697, 2013.
[24] W. Chen, H. Li, Q. Ma and Z. Shang, " Design and implementation of server cluster dynamic load balancing in virtualization environment based on OpenFlow," Proceedings of The Ninth International Conference on Future Internet Technologies. ACM, pp. 1-6, 2014.
[25] W. Chen, Z. Shang, X. Tian, and H. Li, "Dynamic Server Cluster Load Balancing in Virtualization Environment with OpenFlow", Hindawi Publishing Corporation, International Journal of Distributed Sensor Networks, pp. 1-9, 2015.
[26] H. Zhang and X. Guo, " SDN-based load balancing strategy for server cluster," Cloud Computing and Intelligence Systems (CCIS), pp. 1-6, 2014.
[27] M. H. Chen, Y. C. Tien, Y. T. Huang, I. H. Chung, and C. F. Chou, " A Low-Latency Two-Tier Measurement and Control Platform for Commodity SDN," IEEE Communications Magazine, vol. 54.9, pp. 98-104, 2016.
[28] M.F. Bari, S.R. Chowdhury, R. Ahmed and R. Boutaba, "PolicyCop: An Autonomic QoS Policy Enforcement Framework for Software Defined Networks," IEEE SDN for Future Networks and Services (SDN4FNS), pp. 1-7, 2013.
[29] T. Nishio, R. Shinkuma, T. Takahashi and N. B. Mandayam, " Service-oriented heterogeneous resource sharing for optimizing service latency in mobile cloud," Proceedings of the first international workshop on Mobile cloud computing & networking. ACM, pp. 19-26, 2013.
[30] R. Bellman, "On a routing problem", Quart. Appl. Math. 16, pp. 87–90, 1958.
[31] H. W. Kuhn, "The Hungarian method for the assignment problem," Naval research logistics quarterly, vol. 2, pp. 83-97, 1955.
[32] J. Munkres, "Algorithms for the assignment and transportation problems," Journal of the society for industrial and applied mathematics, pp.32-38, 1957.
[33] Python Programming Language Official Website, Available on July 18, 2017: http://www.python.org/.
[34] Munkres implementation for Python. Available on July 14, 2017: https://github.com/bmc/munkres
[35] Z.-J. Yang and C.-S. Yang, "Using Multiple Flow Tables in Software Defined Networking Environment," 2015.
[36] Ryu SDN framework. 2013. Available on June 14, 2017: http://osrg.github.io/ryu
[37] Mininet. Available on June 14, 2017: http://mininet.org/
[38] iPerf, Available on July 18, 2017: https://iperf.fr/