跳到主要內容

簡易檢索 / 詳目顯示

回結果列表
研究生： 黃詣然
Yi-Ran Huang
論文名稱： 資料中心內部路徑之負載平衡
Distributed Packet-Based Load-Balancing for Datacenters
指導教授： 張貴雲
Guey-Yun Chang
口試委員:
學位類別： 碩士
Master
系所名稱： 資訊電機學院 - 資訊工程學系
Department of Computer Science & Information Engineering
論文出版年： 2016
畢業學年度： 104
語文別： 英文
論文頁數： 37
中文關鍵詞： 資料中心 、負載平衡 、分散式
外文關鍵詞： Data center network, Load balancing, Distributed
相關次數： 點閱：9  下載：0 
分享至:
查詢本校圖書館目錄 查詢臺灣博碩士論文知識加值系統 勘誤回報

    • 近年來，負載平衡被認為是評估資料中心一般效能以及計算資源利
    用率的評估標準。現今的方法經常遭遇到路徑負載不平衡以及封包亂
    序的問題。在本篇論文，我們提出一資料中心內部的負載平衡演算法，
    在我們的方法中採用的是基於封包分派的策略，每個葉片交換器在做
    決策的時候是不具備整個資料中心的全局觀，我們方法是利用每個葉
    片交換器曾經轉送過的流量去估算路徑的壅塞程度，而且我們不會修
    改到傳輸層協定。最後，我們利用OMNeT++ 這套模擬軟體去評估我
    們與目前最新穎的方法的效能。

    Load balancing is considered as a measurement to improve general performance
    and computing resource utilization of two-tier data center networks
    (i.e., spine tier and leaf tier). Existing load balancing schemes usually suffer
    from uneven load balancing or packet reordering problem. In this paper,
    we introduce a load balancing algorithm for data center networks. In our
    scheme, every leaf switch makes load balancing decisions without the global
    view about path congestion levels. Each leaf switch leverages traffic volume
    that it has ever sent to calculate the paths’ congestion levels and do not
    require any modification of transport protocol (e.g., packet header modification).
    Final, we use OMNeT++ to evaluate our performance along with the
    state-of-the-art.

    中文􁄔要i Abstract ii 􀤠謝iii Contents iv List of Figures vi List of Tables vii 1 Introduction 1 2 Related Works 3 3 Problem statement 6 3.1 Environment Description . . . . . . . . . . . . . . . . . . . . . . . . . . 6 3.2 Research Goals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7 4 The proposed Algorithm 8 4.1 Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8 4.2 Waiting time in source leaf switch . . . . . . . . . . . . . . . . . . . . . 9 4.2.1 Waiting time for a new arrival packet to depart . . . . . . . . . . 10 4.2.2 Waiting time for a prior packet to depart . . . . . . . . . . . . . . 10 4.2.3 FIFO in source leaf switch . . . . . . . . . . . . . . . . . . . . . 12 4.3 Waiting time in path . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13 4.3.1 Path Load estimation . . . . . . . . . . . . . . . . . . . . . . . . 13 4.4 Path Load Balancing Decision Logic . . . . . . . . . . . . . . . . . . . . 15 5 Simulation 18 5.1 Simulation Environments . . . . . . . . . . . . . . . . . . . . . . . . . . 18 5.1.1 Topologies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18 5.1.2 Traffic Patterns . . . . . . . . . . . . . . . . . . . . . . . . . . . 19 5.1.3 Baseline parameters setting . . . . . . . . . . . . . . . . . . . . . 19 5.2 Simulation Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20 5.2.1 Flow Completion Time . . . . . . . . . . . . . . . . . . . . . . . 20 5.2.2 Load balancing efficiency . . . . . . . . . . . . . . . . . . . . . 20 5.2.3 Packet Reordering . . . . . . . . . . . . . . . . . . . . . . . . . 21 6 Conclusion 22

    [1] M. M. et al. VXLAN, “Vxlan: A framework for overlaying virtualized layer 2 networks
    over layer 3 networks,” 2013.
    [2] “Network virtualization using generic routing encapsulation.”
    [3] B. R. N. H. M. Al-Fares, S. Radhakrishnan and A. Vahdat, “Hedera: dynamic flow
    scheduling for data center networks,” in NSDI, 2010, 2010.
    [4] M. Alizadeh, T. Edsall, S. Dharmapurikar, R. Vaidyanathan, K. Chu, A. Fingerhut,
    V. T. Lam, F. Matus, R. Pan, N. Yadav, and G. Varghese, “Conga: Distributed
    congestion-aware load balancing for datacenters,” in Proceedings of the 2014 ACM
    Conference on SIGCOMM, ser. SIGCOMM ’14. New York, NY, USA: ACM, 2014,
    pp. 503–514. [Online]. Available: http://doi.acm.org/10.1145/2619239.2626316
    [5] H. W. et al, “Cope: Traffic engineering in dynamic networks,” in In SIGCOMM,
    2006, vol. 36, 2006, pp. 99 – 110.
    [6] S. L. A. Elwalid, C. Jin and I. Widjaja, “Mate: Mpls adaptive traffic engineering,”
    in INFOCOM, 2001 Proceedings IEEE, vol. 3, 2001, pp. 1300 – 1309.
    [7] B. D. S. Kandula, D. Katabi and A. Charny, “Walking the tightrope: responsive yet
    stable traffic engineering,” in SIGCOMM, 2005, 2005.
    [8] M. Shreedhar and G. Varghese, “Efficient fair queuing using deficit round robin,”
    1995.
    [9] J. C. et al, “Per-packet load-balanced, low-latency routing for clos-based data center
    networks,” in CoNEXT, 2013, 2013, pp. 49 – 60.
    [10] P. M. D. B. D. Zats, T. Das and R. H. Katz, “Detail: reducing the flow completion
    time tail in datacenter networks,” in SIGCOMM, 2012, vol. 42, 2012, pp. 139 – 150.
    [11] C. R. et al, “Improving datacenter performance and robustness with multipath tcp,”
    in SIGCOMM, 2011, 2011, pp. 266 – 277.
    [12] M. A. et al, “Data center tcp (dctcp),” in SIGCOMM, 2010, 2010.
    [13] R. Gandhi, H. H. Liu, Y. C. Hu, G. Lu, J. Padhye, L. Yuan, and M. Zhang, “Duet:
    Cloud scale load balancing with hardware and software,” in Proceedings of the 2014
    ACM Conference on SIGCOMM, ser. SIGCOMM ’14. New York, NY, USA: ACM,
    2014, pp. 27–38. [Online]. Available: http://doi.acm.org/10.1145/2619239.2626317
    24

    QR CODE
    :::