Achieving faster failure detection in OSPF networks

Abstract

Abstract-With the current default settings of the OSPF parameters, the network takes several tens of seconds before recovering from a failure. The main component in this delay is the time required to detect the failure using Hello protocol. Failure detection time can be speeded up by reducing the value of HelloInterval. However, too small a value of HelloInterval will result in an increased chance of network congestion causing loss of several consecutive Hellos, thus leading to false breakdown of adjacency between routers. Such false alarms not only disrupt network traffic by causing unnecessary routing changes but also increase the processing load on the routers which may potentially lead to routing instability. In this paper, we investigate the following question -What is the optimal value for the HelloInterval that will lead to fast failure detection in the network while keeping the false alarm occurrence within acceptable limits? We examine the impact of both network congestion and the network topology on the optimal HelloInterval value. Additionally, we investigate the effectiveness of faster failure detection in achieving faster failure recovery in OSPF networks. (Abstract)

Overview and Challenges of Overlay Networks: A Survey

“…Goyal et al [65] investigate the impact of topologies and network congestion on optimal HelloInterval for OSPF network through simulation. Basu et al [66] perform experimental study of the stability of OSPF in terms of convergence time, routing load and number of routing flaps.…”

Handling Transient Link Failures Using Alternate Next Hop Counters

“…In [2], analysis shows that the duration of forwarding disruptions must incorporate the times taken to detect the failure, react to it, propagate the failure notification, re-compute forwarding tables, and the time taken to actually install the changes into the Forwarding Information Base (FIB). IPFRR eliminates the time required by all of these except for the failure detection time, which can be shortened by adjusting protocol parameters [3]. This significantly reduces disruption time, as it allows packets to be re-routed via pre-computed backup paths as soon as the failure is detected.…”

Monitoring cycles for fault detection in meshed all-optical networks

“…Since node 6 and 8 are uncovered, link (5,6) and (5,8) are prior to (5,1) . Alternatively to iteration 1, we apply the largest numbered node first rule and select link (5,8) .…”

“…Iteration 2: start from node 5 and there are 3 "uncovered" links: (5,8) and (5,6) (5,1), . Since node 6 and 8 are uncovered, link (5,6) and (5,8) are prior to (5,1) .…”

“…OSPF and IS-IS, also have inherent functionality of fault detection. Some key parameters of OSPF were optimized in [6] to achieve fast fault detection. A joint optical and IP layer method was proposed in [7] to accelerate the detection speed.…”