关键词: 软件定义网络, 流表项超时, 多目标优化, 大象流侦测, 网络性能优化

Abstract: Abstract: Flow entries are forwarding rules generated by controllers in the control plane but stored in the memory of switches and guiding switches in the data plane to process data packets in Software Defined Network (SDN). Every flow entry in switches has timeout that affects the bandwidth cost in SDN control channel, the memory consumption in related switches, and the entire system’s resource management and performance. As most of the existing performance optimization problems in SDN often have a single objective, which does not consider the impact of how flow entries get timed out and the value of the timeout, a multi-objective optimization problem is proposed based on the dynamic timeouts of flow entries to simultaneously minimize the inaccuracy in the detection of elephant flows, the memory cost of flow entries in switches, and the control channel bandwidth occupation. The dynamic timeout combines the hard-timeout and idle-timeout of flow entries, and can be adjusted in a dynamic way. The NSGA-Ⅱ algorithm is used to solve the proposed optimization problem and to analyze the tradeoff among the three objectives. The Bayesian multi-objective optimization algorithm is also applied to improve the quality of the solution set. The results show that the proposed dynamic timeouts can provide a higher detection accuracy, a lower bandwidth occupation, and a smaller switch memory consumption. It significantly improves the overall performance of SDNs.

Key words: Software Defined Network (SDN), flow entry timeout, multi-objective optimization, elephant flow detection, network performance optimization