Rerouting Based Congestion Control in Data Center Networks

Abstract

Data center networks (DCNs) are widely deployed to provide infrastructure backbone for various cloud services. Meanwhile, the emergence of Software-Defined Networking (SDN) makes it easy to monitor the global status of DCNs. In DCNs, traffic bursts or uneven traffic distribution may lead to network congestion, thus deteriorating the overall network performance. Rerouting is an effective method to mitigate network congestion by rerouting flows to paths with sufficient bandwidth. In this paper, we propose an efficient rerouting algorithm by leveraging the SDN technique. We firstly formulate the rerouting problem as a Multi-Commodity Flow (MCF) problem with the objective to minimize the maximum link utilization in the DCN. Since the problem is a NP-hard problem, we propose a heuristic algorithm that jointly considers the current path utilization and the path criticality to approach the optimal solution. Our simulation results demonstrate the superiority of the proposed algorithm in mitigating the network congestion and guaranteeing the performance of the network, over the classical Equal-Cost Multi-Path (ECMP) algorithm and Global First Fit (GFF) algorithm.