AI-assisted WAN link selection for SD-WAN services

1. A method comprising:

receiving, by a software-defined network in a wide area network (SD-WAN) system having a first WAN link and a second WAN link for an SD-WAN service, WAN link characterization data for the first WAN link over a first time period;

based on determining that a current value of a performance metric at a current time has violated a service level agreement (SLA) rule for an application: determining, by a network analysis system of the SD-WAN system and based on processing the WAN link characterization data for the first WAN link using a machine learning model trained with historical WAN link characterization data for one or more WAN links, a predicted value of the performance metric of the first WAN link at a future time; and

providing, by the network analysis system, the predicted value of the performance metric to an SD-WAN edge device of the SD-WAN system to use in determining whether to reassign, based on the predicted value, the application from the first WAN link to the second WAN link.

2. The method of claim 1 , wherein the WAN link characterization data comprises at least one of service data for the first WAN link or performance metric data indicating measured values for the first WAN link over the first time period.

3. The method of claim 1 , wherein determining whether to reassign comprises: comparing the predicted value to the SLA rule; the method further comprising:

reassigning, in response to determining the predicted value violates the SLA rule, the application from the first WAN link to the second WAN link.

4. The method of claim 1 , wherein the WAN link characterization data comprises one or more of a time to first packet, an average length of sessions, and a packet retransmission rate.

5. The method of claim 1 , wherein determining the predicted value of the performance metric of the first WAN link at a future time comprises determining a time interval associated with the predicted value, the method further comprising:

reassigning, in response to determining that the time interval exceeds a tolerance interval, the application from the first WAN link to the second WAN link.

6. The method of claim 5 , wherein the application comprises a first application, wherein the tolerance interval comprises a first tolerance interval associated with the first application, and wherein the first tolerance interval is different from a second tolerance interval associated with a second application.

7. The method of claim 1 , wherein the method further comprises:

determining, by the SD-WAN system based on the WAN link characterization data for the first WAN link over a second time period, one or more periodic time intervals corresponding to a value or range of values of the performance metric; and

reassigning the application from the first WAN link to the second WAN link in response to determining that a current time is within at least one of the one or more periodic time intervals.

8. The method of claim 1 , wherein receiving the WAN link characterization data for the first WAN link comprises receiving the WAN link characterization data from one or more intermediate routers.

9. A software-defined network in a wide area network (SD-WAN) system comprising:

a first WAN link and a second WAN link for an SD-WAN service;

a network analysis system comprising processing circuitry configured to:

receive WAN link characterization data for the first WAN link over a first time period, and

based on a determination that a current value of a performance metric at a current time has violated a service level agreement (SLA) rule for an application: determine based on processing the WAN link characterization data for the first WAN link using a machine learning model trained with historical WAN link characterization data for one or more WAN links, a predicted value of the performance metric of the first WAN link at a future time,

provide the predicted value of the performance metric to an SD-WAN edge device; and

the SD-WAN edge device comprising processing circuitry configured to:

receive the predicted value of the performance metric, and

determine whether to reassign, based on the predicted value of the performance metric, the application from the first WAN link to the second WAN link.

10. The SD-WAN system of claim 9 , wherein the WAN link characterization data comprises one or more of service data for the first WAN link, performance metric data indicating measured values for the first WAN link over the first time period, a time to first packet, an average length of sessions, or a packet retransmission rate.

11. The SD-WAN system of claim 9 , wherein to determine whether to reassign the application from the first WAN link to the second WAN link, the processing circuitry of the SD-WAN edge device is configured to:

compare the predicted value to the SLA rule; and

the processing circuitry of the SD-WAN edge device being further configured to:

reassign, in response to a determination that the predicted value violates the SLA rule, the application from the first WAN link to the second WAN link.

12. The SD-WAN system of claim 9 , wherein to determine the predicted value of the performance metric of the first WAN link at the future time the processing circuitry of the network analysis system is configured to determine a time interval associated with the predicted value,

wherein the processing circuitry of the SD-WAN edge device is configured to determine to reassign, in response to a determination that the time interval exceeds a tolerance interval, the application from the first WAN link to the second WAN link.

13. The SD-WAN system of claim 12 , wherein the application comprises a first application, and wherein the tolerance interval comprises a first tolerance interval associated with the first application, wherein the first tolerance interval is different from a second tolerance interval associated with a second application.

14. The SD-WAN system of claim 9 , further comprising one or more intermediate routers, wherein to receive the WAN link characterization data for the first WAN link, the processing circuitry of the network analysis system is configured to receive the WAN link characterization data from the one or more intermediate routers.

15. Non-transitory computer-readable storage media comprising instructions that, when executed, configure processing circuitry of a software-defined network in a wide area network (SD-WAN) system to:

receive wide area network (WAN) link characterization data for a first WAN link of a SD-WAN service over a first time period;

based on a determination that a current value of a performance metric at a current time has violated a service level agreement (SLA) rule for an application: determine, by a network analysis system of the SD-WAN system and based on processing the WAN link characterization data for the first WAN link using a machine learning model trained with historical WAN link characterization data for one or more WAN links, a predicted value of the performance metric of the first WAN link at a future time; and

provide the predicted value of the performance metric from the network analysis system to an SD-WAN edge device of the SD-WAN system to use in determining whether to reassign, based on the predicted value, the application from the first WAN link to a second WAN link for the SD-WAN service.

16. The non-transitory computer-readable storage media of claim 15 , wherein the WAN link characterization data comprises at least one of service data for the first WAN link over the first time period and/or performance metric data indicating measured values for the first WAN link over the first time period.

17. The non-transitory computer-readable storage media of claim 15 , further comprising instructions that, when executed, configure the processing circuitry of the SD-WAN system to:

compare the predicted value to the SLA rule; and

reassign, in response to a determination that the predicted value violates the SLA rule, the application from the first WAN link to the second WAN link.

18. The non-transitory computer-readable storage media of claim 15 , wherein the WAN link characterization data comprises one or more of a time to first packet, an average length of sessions, and a packet retransmission rate.

19. The non-transitory computer-readable storage media of claim 15 , further comprising instructions that, when executed, configure the processing circuitry of the SD-WAN system to:

determine a time interval associated with the predicted value; and

reassign, in response to a determination that the time interval exceeds a tolerance interval, the application from the first WAN link to the second WAN link.

20. The non-transitory computer-readable storage media of claim 19 , wherein the application comprises a first application, wherein the tolerance interval comprises a first tolerance interval associated with the first application, and wherein the first tolerance interval is different from a second tolerance interval associated with a second application.