Software-defined deployments are growing into data center and enterprise network infrastructures. The typical promises of software-defined networks (SDN) are improved time for market, decreased risk and operational costs for services, flexibility and unified management. However, little is known and shared about how to actually manage an SDN network, especially in localising underperforming network paths (what we call “troubleshooting”).
We describe a novel approach to ease large network troubleshooting by leveraging SDN features and incorporating distributed monitoring of network traffic. We suggest SDN-RADAR, a tool that can help network administrators understand which is the most likely faulty network link. To the best of our knowledge this is the first troubleshooting solution that combines user-side performance measurements with network data extracted from the SDN controller.