Traffic Modeling Characterization of Software Defined Networks

Software Defined Networking (SDN) is a trending technology and its popularity can be attributed to its speed and agility in deploying new applications. Realistic modeling of SDN traffic is crucial in designing control algorithms and in provisioning network resources. The purpose of this paper is to model and characterize the SDN traffic. To this end, we have created an SDN environment using the mininet emulator and OpenDayLight (ODL) controller using which we have collected the SDN traffic data traces. The quantiles of inter-arrival times of SDN data traffic have been plotted against the quantiles of theoretical distributions. Then the Kolmogorov Smirnov (KS) test has been applied to model the SDN traffic and to capture its distribution characteristics effectively. From the KS test, it has been inferred that the inter arrival times of SDN traffic follows the Generalized Extreme Value (GEV) distribution. These findings indicate that traditional Poisson models for SDN traffic may not be suitable for evaluating the performance of Software Defined Networks.