A Scalable Traffic Engineering Method in an SDN-based Data Center Network Using Decomposition Technique

Today's data centers consist of thousands of servers hosting a variety of cloud-based services. In this paper, a scalable and new technique is presented for traffic engineering in software-defined data center networks helping to design an optimal demand-path mapping with a tolerable computational complexity. The proposed method is based on linear programming model and attempts to minimize the maximum link utilization that causes minimum link congestion. The method focuses on an optimal solution to load balancing in networks. It proposes a new decomposition technique that can limit the search space of the original Linear Programming problem, such that the time required to solve the problem can be reduced substantially. To reduce time complexity, a decomposition technique is used to divide the problem model into smaller sub-problems, which are then solved simultaneously by applying parallelizing techniques (multiple-core computing and OpenMP).Simulation results show that solving time and load balancing are considerably improved.