【摘要】 由于网络流量动态变化,控制器负载均衡成为大规模部署软件定义网络研究的重点。提出基于Q-learning的动态交换机迁移算法,首先对软件定义网络中的控制器部署问题建模,再应用Q-learning反馈机制学习实时网络流量,最后根据Q表格将交换机从高负载控制器动态迁移到低负载控制器上,实现控制器的负载均衡。仿真结果表明,所提算法能够获得较低的控制器负载标准方差。更多还原

【Abstract】 Load balance for controllers becomes an important research issue for large- scale deployed Software Defined Networks( SDN) with the dynamic network load. A dynamic switch migration algorithm based on Q-learning is proposed in this paper,which models controller placement problem firstly,then uses feedback scheme of Q-learning to learn the real time network load,and migrates switches from high-load controllers to low-load controllers on the bias of Q table lastly,to realize load balance of controllers. Simulation results show that the proposed algorithm gets low standard deviation of load born on controllers.更多还原