Dynamic Risk-Aware Routing for OSPF networks

Demand for reliable service requires network operators to provide Quality of Service (QoS) and availability always more important to remain competitive. Since availability is not an option, they must deal with incidents using mechanisms such as the protection and restoration. Protection can restore a path extremely quickly, but is very costly in terms of resources, while restoration is much cheaper, but generates a longer outage. Moreover, even if performances are different, all these mechanisms are reactive and therefore can not completely mask the effects of a failure for end users. This paper focuses on the IP restoration mechanism provided by the Interior Gateway Protocol (IGP) such that Open Shortest Path First (OSPF) and Intermediate system to intermediate system (IS-IS). The time between the occurrence of a failure and the end of the OSPF convergence implies packet loss, temporary routing loops or routing black holes that disrupt the operator's routing topology. Indeed, the recovery mechanism is only involved after the failure occurrence which only allows to limit the failure impact on traffic, but not remove it completely. We propose a complementary approach that eliminates this default by anticipating failures with a Risk Assessment Module (RAM), in order to override the slowness of recovery mechanism. This allows to completely eliminate the failure incidence on traffic by performing reconfiguration actions few seconds before the failure. The failure prediction is exploiting to provide a Risk-