Scoring policies for predictive routing suggestions

1. A method comprising:

making, by a device, a determination that a first predictive routing policy generated by a predictive routing engine for a network would have performed better than a preexisting routing policy that is active in the network;

adjusting, by the device and based on the determination, a level of trust associated with the predictive routing engine;

obtaining, by the device, information regarding a second predictive routing policy generated by the predictive routing engine for the network; and

activating, by the device, the second predictive routing policy in the network, based on the level of trust associated with the predictive routing engine.

2. The method as in claim 1 , wherein the network comprises a software-defined wide area network (SD-WAN).

3. The method as in claim 1 , wherein adjusting the level of trust associated with the predictive routing engine comprises:

increasing the level of trust from a level that prevents activation of any predictive routing policies generated by the predictive routing engine in the network to a level that allows automatic activation of at least some predictive routing policies generated by the predictive routing engine.

4. The method as in claim 1 , further comprising:

rolling back activation of the second predictive routing policy in the network, based on a determination that the second predictive routing policy degrades performance in the network.

5. The method as in claim 1 , wherein making the determination that the first predictive routing policy generated by the predictive routing engine for the network would have performed better than the preexisting routing policy that is active in the network comprises:

determining that a service level agreement violation occurred as a result of the preexisting routing policy; and

determining that the service level agreement violation would not have occurred had the first predictive routing policy been active in lieu of the preexisting routing policy.

6. The method as in claim 1 , wherein activating the second predictive routing policy in the network comprises:

determining whether a temporal conflict exists between the second predictive routing policy and any routing policies already active in the network.

7. The method as in claim 6 , wherein activating the second predictive routing policy in the network comprises:

resolving a temporal conflict between the second predictive routing policy and a particular routing policy, based on the second predictive routing policy having a higher confidence score than that of the particular routing policy, by activating the particular routing policy only at times outside of a time period during which the second predictive routing policy is active.

8. The method as in claim 1 , wherein activating the second predictive routing policy in the network comprises:

determining whether the second predictive routing policy would override a non-overridable routing policy already active in the network.

9. The method as in claim 1 , wherein activating the second predictive routing policy in the network, based on the level of trust associated with the predictive routing engine, further comprises:

receiving, via a user interface, an instruction to activate the second predictive routing policy in the network during a trial period, wherein the second predictive routing policy is activated temporarily during the trial period.

10. The method as in claim 1 , wherein the preexisting routing policy is a user-defined or default routing policy.

11. An apparatus, comprising:

one or more network interfaces;

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:

make a determination that a first predictive routing policy generated by a predictive routing engine for a network would have performed better than a preexisting routing policy that is active in the network;

adjust, based on the determination, a level of trust associated with the predictive routing engine;

obtain information regarding a second predictive routing policy generated by the predictive routing engine for the network; and

activate the second predictive routing policy in the network, based on the level of trust associated with the predictive routing engine.

12. The apparatus as in claim 11 , wherein the network comprises a software-defined wide area network (SD-WAN).

13. The apparatus as in claim 11 , wherein the apparatus adjusts the level of trust associated with the predictive routing engine by:

increasing the level of trust from a level that prevents activation of any predictive routing policies generated by the predictive routing engine in the network to a level that allows automatic activation of at least some predictive routing policies generated by the predictive routing engine.

14. The apparatus as in claim 11 , wherein the process when executed is further configured to:

roll back activation of the second predictive routing policy in the network, based on a determination that the second predictive routing policy degrades performance in the network.

15. The apparatus as in claim 11 , wherein the apparatus makes the determination that the first predictive routing policy generated by the predictive routing engine for the network would have performed better than the preexisting routing policy that is active in the network by:

determining that a service level agreement violation occurred as a result of the preexisting routing policy; and

determining that the service level agreement violation would not have occurred had the first predictive routing policy been active in lieu of the preexisting routing policy.

16. The apparatus as in claim 11 , wherein the apparatus activates the second predictive routing policy in the network by:

determining whether a temporal conflict exists between the second predictive routing policy and any routing policies already active in the network.

17. The apparatus as in claim 16 , wherein the apparatus actives the second predictive routing policy in the network by:

resolving a temporal conflict between the second predictive routing policy and a particular routing policy, based on the second predictive routing policy having a higher confidence score than that of the particular routing policy, by activating the particular routing policy only at times outside of a time period during which the second predictive routing policy is active.

18. The apparatus as in claim 11 , wherein the apparatus activates the second predictive routing policy in the network by:

determining whether the second predictive routing policy would override a non-overridable routing policy already active in the network.

19. The apparatus as in claim 11 , wherein the preexisting routing policy is a user-defined or default routing policy.

20. A tangible, non-transitory, computer-readable medium storing program instructions that cause a device to execute a process comprising:

making, by the device, a determination that a first predictive routing policy generated by a predictive routing engine for a network would have performed better than a preexisting routing policy that is active in the network;

adjusting, by the device and based on the determination, a level of trust associated with the predictive routing engine;

obtaining, by the device, information regarding a second predictive routing policy generated by the predictive routing engine for the network; and

activating, by the device, the second predictive routing policy in the network, based on the level of trust associated with the predictive routing engine.