Nested Failure Detection and Recovery in Software Defined Networks

Deepa B Swarna, V. Muthumanikandan · 2019

considering the increasing operating costs for the network equipment and the growing demands of the modern data center, switching from traditional methods and protocols to more open and innovation friendly approach of SDN is important. Although Open flow switches can identify link failures in SDN networks, since it does not have control plane in itself, it has to wait for the controller to suggest any alternate routes to respond. Hence path protection and network recovery from failure is crucial in SDN. Service providers would be able to meet their end users expectation only when there is good network quality. Currently in SDN networks, link failures are addressed by rerouting traffic from the switch where link breaks. In the scenario where group based restoration is employed by grouping all disrupted flows at the source of the link failure using a virtual LAN tag, any nested failure in the restored path may result in packet drop at the nodes connected to the nested link failure. Hence an extensive approach should be adopted to address the nested link failures, by including nested failure detection logic in all intermediate switches with respect to the first link failure, so that the packets traversing the subsequent nested failed link can be restored and re-routed effectively, without incurring any packet loss at the source and destination of subsequent nested link failures. The objective of this work is to detect nested link failures by triggering detection logic at the switches between source and destination of first link failure. Whenever a particular link fails, the objective is to examine if it is a nested failure. If the link failure detected is a nested failure, corrective action is taken at the source and destination of the failed switch by grouping all disrupted flows using a new virtual network by matching on the virtual network group id of previous link failure. This would help in eliminating the packet loss that would otherwise be incurred for such links. Subsequently, the network throughput is increased substantially during nested failures.

Read the paper · More papers on PaperTik