Controller robust placement with dynamic traffic in software-defined networking

For dynamic traffic of switches in the software-defined networking, the existing methods mostly adopt the switch migration to balance the controller with an uneven load. However, these researches have overlooked that the frequent migration of switches will cause instability in the control plane and increase the network delay. To reduce switch migration, this paper proposes a robust controller placement model that considers dynamic traffic while placing the controllers. Firstly, the bounded symmetric interval is used to describe the dynamic traffic in the network, and the magnitude of the traffic change is represented by the variation parameter σ. Secondly, the control parameter Γ is introduced to control the conservativeness of the solution and construct a multi-controller robust placement model. Finally, the model is transformed into a linear programming problem based on the strong duality theory, and we use the tabu search algorithm to implement an efficient solution. The simulation results show that different controller placement strategies can be obtained by changing the control parameter. When the service level is low, the cost increase can greatly improve the service level; however, when the service level exceeds 90%, the 10% increase in placement cost can only improve the service level by 1%.