一种可扩展的软件定义数据中心网络流调度策略

A Scalable Traffic Scheduling Policy for Software Defined Data Center Network

针对目前大规模软件定义数据中心网络流量路由机制中可扩展性低所带来的性能瓶颈问题,该文提出一种面向树型结构的数据中心网络分段路由机制。该机制利用边缘交换机对数据流进行阈值检测以区分大小流,同时为满足其不同业务的QoS保证和网络可扩展性的要求,该文提出一种针对大流的在线最先适应算法。最后,利用Mininet在Fat-tree结构上进行实验仿真验证,仿真结果表明,与传统的ECMP算法和Mahout算法相比,该机制在降低了控制器总开销的同时还提高了网络吞吐率。

According to the low scalability of current large-scale software defined data center network traffic routing mechanism which causes network performance bottleneck, this paper proposes a data center network Segment Routing （SR） mechanism based on OpenFlow. The mechanism distinguishes the size of the flow by making use of edge switches to conduct a data stream threshold test. In order to meet the QoS guarantee and network scalability requirements of different services, this paper proposes a segment routing algorithm for elephant flow.Finally, Mininet is utilized for experiment simulation on Fat-tree topology. Compared to the traditional ECMP algorithm and Mahout algorithm, simulation results show that the mechanism reduces the overhead of controller, and has better network throughout.