Towards accurate online traffic matrix estimation in software-defined networks

Abstract

Traffic matrix measurement provides essential information for network design, operation and management. In today's networks, it is challenging to get accurate and timely traffic matrix due to the hard resource constraints of network devices. Recently, Software-Defined Networking (SDN) technique enables customizable traffic measurement, which can provide flexible and fine-grain visibility into network traffic. However, the existing software-defined traffic measurement solutions often suffer from feasibility and scalability issues. In this paper, we seek accurate, feasible and scalable traffic matrix estimation approaches. We propose two strategies, called Maximum Load Rule First (MLRF) and Large Flow First (LFF), to design feasible traffic measurement rules that can be installed in TCAM entries of SDN switches. The statistics of the measurement rules are collected by the controller to estimate fine-grained traffic matrix. Both MLRF and LFF satisfy the flow aggregation constraints (determined by associated routing policies) and have low-complexity. Extensive simulation results on real network and traffic traces reveal that MLRF and LFF can achieve high accuracy of traffic matrix estimation and high probability of heavy hitter detection.