摘要 网络拥塞会严重影响网络运行的性能和稳定性．Openflow网络中控制器可以对网络进行集中式管理，基于这种特性，提出一种与业务属性相关的OpenFlow网络拥塞管理机制(Congestion Management Related Traffic Attribution in OpenFlow，CMTAOF)，以实现更高效的拥塞控制．CMTAOF机制利用控制器掌握全局网络流量分布的优势，将数据流的业务属性、网络的流量分布和网络的拓扑结构相结合，在拥塞链路处，选择一条或者多条转移开销最小的数据流从网络入口进行重路由．理论分析和实验表明，和传统的拥塞分流策略以及现有的基于OpenFlow网络的LABERIO机制相比，CMTAOF既能够有效的缓解网络的拥塞状况，保证用户的服务质量，同时还能够提升网络的吞吐量和资源利用率．
Abstract：Network performance and stability will be heavily affected by congestion.Recently,a new network framework named OpenFlow is proposed in which the network can be centrally managed.Take advantage of this characteristic,to achieve higher performance,a new congestion control mechanism is presented in this paper,that is CMTA-OF(Congestion Management Based on Traffic Attribution in OpenFlow).The main idea is that controller observes traffic distribution and network topology globally.When congestion occurs,flows with minimum transfer cost will be rerouted when entering OpenFlow network.Analysis and experiment proves that CMTA-OF can realize congestion,ensure QoS and improve throughput and resource utilization.
2015, Vol. 36 
