Improving network convergence time and network stability of an OSPF-routed IP networ

Loss of routing protocol control messages due to network congestion can cause peer failures in routers, leading to routing failures and network instability. We investigate the effect of network congestion on Open Shortest Path First (OSPF) routing protocol, when OSPF control traffic is not isolated from the data traffic. We attempt to decrease the network convergence time in response to a node or link failure in a network. We develop an algorithm to dynamically tune OSPF failure threshold based on the network congestion level. The tunable OSPF protocol expedites failure detections as compared to standard OSPF protocol in lightly loaded networks by reducing the failure threshold parameter; whereas the failure threshold is increased during network congestion to sustain stable routing operation and avoid spurious failure decisions due to missed or delayed hello packets. Our findings demonstrate the importance of tunable failure threshold to achieve stable and robust network operation.