Partitioning Traffic Engineering in Software Defined Wide Area Networks
Yufeng Xin, Yifei Wang · 2023
Traffic engineering (TE) is experiencing a surge in research and development interests due to the rise of software-defined networking and the massive increase in data volume in wide area networks. State-of-the-art TE systems address computational complexity and resiliency challenges in centralized controllers through aggregation or partitioning of the underlying network topology. This paper introduces a new TE load balancing method within a novel scalable SD-WAN TE controller framework that partitions traffic demands into multiple groups. Based on an optimization formulation with bandwidth and latency constraints, we show that effective partitioning algorithms allow solving a set of smaller optimization problems with much reduced computation time. Numerical analysis validates the viability of the proposed approach and demonstrates that a small number of traffic groups can achieve the desired TE performance.