Flow Characteristics Aware Dynamic Controller Assignment in Software-defined Networking
-
摘要: 在部署分布式控制平面的软件定义网络中，控制器与交换机的关联仅以数据流请求的数量分布作为控制资源分配的依据。该文针对这一问题，以数据流的源目的地址特征为例，对不同特征数据流的控制资源消耗进行了分析，提出在控制资源分配中应对数据流的特征分布加以考虑。然后，设计了一种流特征感知的控制器关联决策机制，并针对网络流的动态变化特性设计了一种快速求解算法。仿真结果表明，与基于负载均衡的机制对比，所提机制在使用模拟退火算法求解时能节省10%～20%的控制资源消耗；所提快速求解算法可节省10%的资源消耗，且相比模拟退火算法具有较大的速度优势和良好的可扩展性。Abstract: In Software-Defined Networking (SDN) with distributed control plane, the switches are assigned to controllers using only the quantity distribution of flow requests as the basis of resource allocation. To address this issue, the control resource consumption of flow requests processing with different characteristics is analyzed taking the source and destination of flow as an example, from which a conclusion is drawn that the characteristics distribution of flow should be taken into account when allocating control resource. Then, a flow characteristics aware controller assignment model is designed, and a fast algorithm coping with the fluctuation of flow request is proposed. Simulation results show that when solving with the simulated annealing algorithm, the model can save 10%～20% of control resource compared with the load balancing model; with 10% of resource saving, the proposed algorithm outperforms the simulated annealing algorithm in execution speed and scalability.
-
表 1 排序关联算法
算法1 排序关联算法 输入：交换机间流请求速率矩阵 ${{F}} = {[{f_{ij}}]_{N \times N}}$; 控制域间资源消耗矩阵 ${P} = {[{P_{hk}}]_{M \times M}}$; 控制器容量 ${a_m}$; 控制器与交换机之间的距离阈值 $\delta $; 输出：控制器与交换机的关联关系 $\varphi ( \cdot )$ (1) 对每个交换机，根据距离约束公式式(3)确定候选控制器集合
$\varGamma ({s_i})$;(2) 将交换机对 $({s_i},{s_j})$按 ${f_{ij}}$降序排列，形成序列 ${L_S}$; (3) 初始化已关联交换机集合 $\varOmega = \varnothing $; (4) for $({s_i},{s_j}) \in {L_S}$ do (5) if ${s_i} \notin \varOmega $ or ${s_j} \notin \varOmega $ then /*若2个交换机均已关联，则跳过*/ (6) 获取 $\varGamma ({s_i})$与 $\varGamma ({s_j})$, $({s_i},{s_j})$的候选控制器对 $({c_h},{c_k})$的集合为
直积 $\varGamma ({s_i}) \times \varGamma ({s_j})$;(7) 将候选控制器对按 ${p_{hk}}$升序排列，形成序列 $L_C^{ij}$; (8) for $({c_h},{c_k}) \in L_C^{ij}$ do (9) 将 $({s_i},{s_j})$关联到 $({c_h},{c_k})$; (10) if ${\theta _h} < {a_h}$ and ${\theta _k} < {a_k}$ then /*若关联后有控制器超
载，则跳过*/(11) $\varphi ({s_i}) = {c_h}$, $\varphi ({s_j}) = {c_k}$; /*控制器接受关联*/ (12) $\varOmega = \varOmega \cup \{ {s_i},{s_j}\} $; (13) $\varGamma ({s_i}) = \{ \varphi ({s_i})\} $, $\varGamma ({s_j}) = \{ \varphi ({s_j})\} $; (14) goto step 4; (15) end if (16) end for (17) end if (18) end for 
下载: 导出CSV
表 3 实验拓扑数据
拓扑 节点数 边数 控制器个数 控制器容量 距离阈值 RedIris 19 32 4 600 3 GEANT2009 34 52 6 1800 3 DFN 58 87 8 3000 4 Interoute 110 159 10 8000 5 TATANld 145 194 12 12000 7
下载: 导出CSV
表 4 算法运行时间(ms)
拓扑 SeqAsn SimAnn RedIris 6.5 78.36 GEANT2009 8.9 449.68 DFN 11.68 841.06 Interoute 18.32 2223.36 TATANld 25.84 4355.86
下载: 导出CSV
-
KREUTZ D, RAMOS F M V, VERISSIMO P E, et al. Software-defined networking: A comprehensive survey[J]. Proceedings of the IEEE, 2015, 103(1): 14–76 doi: 10.1109/JPROC.2014.2371999 KARAKUS M and DURRESI A. A survey: Control plane scalability issues and approaches in Software-Defined Networking (SDN)[J]. Computer Networks, 2017, 112: 279–293 doi: 10.1016/j.comnet.2016.11.017 BERDE P, GEROLA M, HART J, et al. ONOS: towards an open, distributed SDN OS[C]. Hot Topics in Software Defined Networking (HotSDN), Chicago, USA, 2014: 1–6. LIAO Jianxin, SUN Haifeng, WANG Jingyu, et al. Density cluster based approach for controller placement problem in large-scale software defined networkings[J]. Computer Networks, 2017, 112: 24–35 doi: 10.1016/j.comnet.2016.10.014 DIXIT A, HAO F, MUKHERJEE S, et al. Towards an elastic distributed SDN controller[J]. ACM SIGCOMM Computer Communication Review, 2013, 43(4): 7–12 doi: 10.1145/2534169.2491193 JIMENEZ Y, CERVELLO-PASTOR C, and Garcia A J. On the controller placement for designing a distributed SDN control layer[C]. IFIP Networking Conference, Trondheim, Norway, 2014: 1–9. 伊鹏, 刘邦舟, 王文博, 等. 一种考虑软件定义网络控制节点故障的控制器部署和交换机迁移方法[J]. 电子与信息学报, 2017, 39(8): 1972–1978 doi: 10.11999/JEIT161216YI Peng, LIU Bangzhou, WANG Wenbo, et al. Controller placement and switch immigration strategy for SDN controller failure[J]. Journal of Electronics&Information Technology, 2017, 39(8): 1972–1978 doi: 10.11999/JEIT161216 BENSON T, AKELLA A, and MALTZ D. Network traffic characteristics of data centers in the wild[C]. ACM Internet Measurement Conference (IMC), Melbourne, Australia, 2010: 267–280. MCKEOWN N, ANDERSON T, BALAKRISHNAN H, et al. OpenFlow: Enabling innovation in campus networks[J]. ACM SIGCOMM Computer Communication Review, 2008, 38(2): 69–74 doi: 10.1145/1355734.1355746 GAO Xiaofeng, KONG Linghe, LI Weichen, et al. Traffic load balancing schemes for devolved controllers in mega data centers[J]. IEEE Transactions on Parallel and Distributed Systems, 2017, 28(2): 572–585 doi: 10.1109/TPDS.2016.2579622 WANG Tao, LIU Fangming, GUO Jian, et al. Dynamic SDN controller assignment in data center networks: Stable matching with transfers[C]. IEEE International Conference on Computer Communications (INFOCOM), San Francisco, USA, 2016: 1–9. WANG Tao, LIU Fangming, and XU Hong. An efficient online algorithm for dynamic SDN controller assignment in data center networks[J]. IEEE/ ACM Transactions on Networking, 2017, 25(5): 2788–2801 doi: 10.1109/TNET.2017.2711641 SOLIMAN M, NANDY B, LAMBADARIS I, et al. Source routed forwarding with software defined control, considerations and implications[C]. International Conference on emerging Networking EXperiments & Technologies (CoNEXT), Nice, France, 2012: 43–44. LOIOLA E M, ABREU N M M D, BOAVENTURA-NETTO P O, et al. A survey for the quadratic assignment problem[J]. European Journal of Operational Research, 2007, 176(2): 657–690 doi: 10.1016/j.ejor.2005.09.032 KIRKPATRICK S. Optimization by simulated annealing: quantitative studies[J]. Journal of Statistical Physics, 1984, 34(5): 975–986 doi: 10.1007/BF01009452 HARDY G H, LITTLEWOOD J E, and PÓLYA G. Inequalities[M]. New York: Cambridge University Press, 1952: 261–262. -
图(4) / 表(4)
计量
- 文章访问数: 2317
- HTML全文浏览量: 469
- PDF下载量: 47
- 被引次数: 0
下载:
