In a SDN solution it is crucial to have a self healing network that detects node and link failure events to recover the connectivity of a system by exploiting redundant links; all of which is transparent to the user. Failover mechanisms can be implemented either using slow-reroute or hard-reroute. 1. Slow-reroute is done in software using disjoint path calculation algorithms (E.g. Suurballe algorithm) and standard Openflow features. 2. Fast-reroute is done using hardware assisted OAM with 802.1ag path failure detection and using Openflow group failover table features. Opendaylight controller currently implements path failure detection and recovery using slow-reroute mechanism. In this presentation we discuss the current implementation of failover mechanisms, and the challenges in supporting it in the SDN paradigm.
