A Rapid Fault Detection Method for SoftwareDefined Network

With the rapid growth of modern network scales, the timeliness of topology information is critical for network management and resource allocation. If topology information updates are delayed, faults cannot be detected in time for traffic rerouting, leading to prolonged service interruptions and negatively affecting user experience. Although existing fast detection mechanisms such as BFD, FRR, and heartbeat detection can reduce fault detection time, those mechanisms still face limitations in large-scale network. In large-scale networks, the SDN controller is a highly effective management tool that enables centralized control, dynamic traffic adjustment, rapid fault recovery, and simplified network expansion. The controller typically uses the BGP-LS protocol to collect network topology. However, practical deployments have revealed that when devices become disconnected, the controller relies on the timeout mechanism of network protocols to detect faults, resulting in slow updates of device states and reduced network responsiveness. To address this issue, this study proposes a rapid detection mechanism based on link-down events. By analyzing link-down events in real time, this method quickly determines the disconnection status of related nodes and updates topology information immediately upon confirmation. Experimental results show that this mechanism significantly reduces device disconnection detection time and has minimal impact on controller resource requirements, demonstrating excellent scalability and practicality. It provides an efficient and reliable solution for network topology management in SDN environments.