An efficient latency monitoring scheme in software defined networks

Abstract The new architecture caught the attention of more and more scholars, called Software Defined Networks (SDN), separates the control plane from the data plan. Meanwhile, many applications which are sensitive to the QoS demand the network reach a certain latency constraint. The SDN brings new opportunity for measuring latency. However, with the expansion of network scale and the requirement of time-delay accuracy, the performance of the controller itself becomes the bottleneck of the requirements. This paper is based on the SDN latency monitoring solution to improve the efficiency and performance of the detection. Periodical triangular function is used in this paper to fit the available latent data so as to judge the variation period and more accurately fit the actual latency data. An optimization algorithm for detection period and time is proposed in this paper which can reduce the detection and time while ensuring the accurate latency monitoring. Compared to other method, the experimental results indicate that it allows for more accurate delay monitoring with smaller performance and network resources in the proposed algorithm.