Predictive fast reroute
In one embodiment, a router in a network reports, a supervisor, capabilities of the router to support fast reroute. The router receives a prediction model from the supervisor that is able to predict failures along a path in the network. The router predicts, using the prediction model, a failure along a primary path in the network that is currently being used by the router to send traffic. The router performs, in advance of the failure predicted by the router, a fast reroute of at least a portion of the traffic from the primary path to a backup path in the network.
1 . A method comprising:
reporting, by a router in a network and to a supervisor, capabilities of the router to support fast reroute and local state information for a backup path for a primary path used by the router, the local state information including whether the backup path provides a bandwidth guarantee;
receiving, at the router, a prediction model from the supervisor that is able to predict failures along a path in the network, the prediction model having been trained by the supervisor using the local state information reported by the router;
executing, by the router, the prediction model to predict a failure along the primary path in the network that is currently being used by the router to send traffic and to generate a fast reroute recommendation indicating whether to fast reroute at least a portion of the traffic from the primary path to the backup path based on a prediction by the prediction model of whether traffic rerouted from the primary path to the backup path would experience a failure or service level agreement violation on the backup path; and
responsive to the fast reroute recommendation, performing, by the router and in advance of the failure predicted by the router, a fast reroute of at least a portion of the traffic from the primary path to the backup path in the network.
2 . The method as in claim 1 , wherein the router is a P router or a provider edge (PE) router.
3 . The method as in claim 1 , wherein the router uses Multiprotocol Label Switching Fast Reroute (MPLS FRR) to perform the fast reroute.
4 . The method as in claim 1 , wherein the router uses Internet Protocol (IP) Fast Reroute (IP FRR) to perform the fast reroute.
5 . The method as in claim 1 , further comprising:
obtaining, by the router, pre-reroute telemetry by probing the backup path in response to predicting the failure; and
obtaining, by the router, post-reroute telemetry by continuing to probe the backup path after performing the fast reroute to the backup path.
6 . The method as in claim 5 , further comprising:
determining a performance of the prediction model based on the pre-reroute telemetry and on the post-reroute telemetry.
7 . The method as in claim 1 , wherein the failure predicted by the router is a predicted violation of a service level agreement (SLA) by the primary path.
8 . The method as in claim 1 , further comprising:
selecting, by the router, the portion of the traffic to be fast rerouted onto the backup path based on one or more applications associated with it.
9 . The method as in claim 1 , wherein the capabilities of the router to support fast reroute are reported by the router to the supervisor via an Open Shortest Path First (OSPF) or Intermediate-System-to-Intermediate-System (ISIS) message.
10 . The method as in claim 1 , wherein the backup path is a tunnel in the network pre-established by the router before predicting the failure of the primary path.
11 . An apparatus, comprising:
one or more network interfaces to communicate with a network;
a processor coupled to the one or more network interfaces and configured to execute one or more processes; and
a memory configured to store a process that is executable by the processor, the process when executed configured to:
report to a supervisor, capabilities of the apparatus to support fast reroute and local state information for a backup path for a primary path used by the apparatus, the local state information including whether the backup path provides a bandwidth guarantee;
receive, at the apparatus, a prediction model from the supervisor that is able to predict failures along a path in the network, the prediction model having been trained by the supervisor using the local state information reported by the apparatus;
execute, at the apparatus, the prediction model to predict a failure along the primary path in the network that is currently being used by the apparatus to send traffic and to generate a fast reroute recommendation indicating whether to fast reroute at least a portion of the traffic from the primary path to the backup path based on a prediction by the prediction model of whether traffic rerouted from the primary path to the backup path would experience a failure or service level agreement violation on the backup path; and
perform, in advance of the failure predicted by the apparatus, a fast reroute of at least a portion of the traffic from the primary path to the backup path in the network.
12 . The apparatus as in claim 11 , wherein the apparatus is a P router or a provider edge (PE) router.
13 . The apparatus as in claim 11 , wherein the apparatus uses Multiprotocol Label Switching Fast Reroute (MPLS FRR) to perform the fast reroute.
14 . The apparatus as in claim 11 , wherein the apparatus uses Internet Protocol (IP) Fast Reroute (IP FRR) to perform the fast reroute.
15 . The apparatus as in claim 11 , wherein the process when executed is further configured to:
obtain pre-reroute telemetry by probing the backup path in response to predicting the failure; and
obtain post-reroute telemetry by continuing to probe the backup path after performing the fast reroute to the backup path.
16 . The apparatus as in claim 15 , wherein the process when executed is further configured to:
determine a performance of the prediction model based on the pre-reroute telemetry and on the post-reroute telemetry.
17 . The apparatus as in claim 11 , wherein the failure predicted by the apparatus is a predicted violation of a service level agreement (SLA) by the primary path.
18 . The apparatus as in claim 11 , wherein the process when executed is further configured to:
select the portion of the traffic to be fast rerouted onto the backup path based on one or more applications associated with it.
19 . The apparatus as in claim 11 , wherein the capabilities of the apparatus to support fast reroute are reported by the apparatus to the supervisor via a routing protocol message.
20 . A tangible, non-transitory, computer-readable medium storing program instructions that cause a router in a network to execute a process comprising:
reporting, by the router and to a supervisor, capabilities of the router to support fast reroute and local state information for a backup path for a primary path used by the router, the local state information including whether the backup path provides a bandwidth guarantee;
receiving, at the router, a prediction model from the supervisor that is able to predict failures along a path in the network, the prediction model having been trained by the supervisor using the local state information reported by the router;
executing, by the router, the prediction model to predict a failure along the primary path in the network that is currently being used by the router to send traffic and to generate a fast reroute recommendation indicating whether to fast reroute at least a portion of the traffic from the primary path to the backup path based on a prediction by the prediction model of whether traffic rerouted from the primary path to the backup path would experience a failure or service level agreement violation on the backup path; and
responsive to the fast reroute recommendation, performing, by the router and in advance of the failure predicted by the router, a fast reroute of at least a portion of the traffic from the primary path to the backup path in the network.