Intelligent tunnel assignment and dynamic SLA threshold configuration to increase availability and utilization of SD-WAN tunnels

1. A method comprising:

tracking, by a supervisory service for a software-defined wide area network (SD-WAN), a performance metric for a tunnel in the SD-WAN;

computing, by the supervisory service, a cumulative distribution function (CDF) for the tracked performance metric;

assessing, by the supervisory service, curvature of the CDF for the tracked performance metric relative to a service level agreement (SLA) threshold of an application for that performance metric; and

controlling, by the supervisory service, assignment of traffic for the application to the tunnel, based on the assessed curvature of the CDF for the tracked performance metric relative to the SLA threshold of the application for that performance metric,

wherein the assessing of the curvature of the CDF for the tracked performance metric relative to the SLA threshold of the application for that performance metric comprises:

computing, by the supervisory service, a point of slope increase on the CDF; and

determining, by the supervisory service, a distance between the computed point of slope increase on the CDF and the SLA threshold of the application for that performance metric, wherein the computed point of slope increase on the CDF is either above or below the SLA threshold.

2. The method as in claim 1 , wherein controlling assignment of the traffic for the application to the tunnel, based on the assessed curvature of the CDF for the tracked performance metric relative to the SLA threshold of the application for that performance metric, comprises:

determining that the SLA threshold is below the point of slope increase on the CDF beyond a predefined distance; and

assigning the traffic for the application to the tunnel, based on the determination.

3. The method as in claim 1 , wherein controlling assignment of the traffic for the application to the tunnel, based on the assessed curvature of the CDF for the tracked performance metric relative to the SLA threshold of the application for that performance metric, comprises:

determining that the SLA threshold is above the point of slope increase on the CDF beyond a predefined distance; and

prohibiting the traffic for the application from being assigned to the tunnel, based on the determination.

4. The method as in claim 1 , wherein controlling assignment of the traffic for the application to the tunnel, based on the assessed curvature of the CDF for the tracked performance metric relative to the SLA threshold of the application for that performance metric, comprises:

determining that the SLA threshold is near the point of slope increase on the CDF when the distance between the point of slope increase on the CDF and the SLA threshold is less than a predefined distance; and

flagging the tunnel as SLA threshold sensitive, based on the determination.

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

adjusting the SLA threshold of the application, in an effort to make the tunnel non-SLA threshold sensitive; and

assigning the traffic for the application to the tunnel, with the adjusted SLA threshold.

6. The method as in claim 5 , further comprising:

monitoring a performance metric for the application, after adjusting the SLA threshold of the application; and

dynamically re-adjusting the SLA threshold of the application, when the performance metric for the application is unacceptable.

7. The method as in claim 5 , further comprising:

monitoring a performance metric for the application, after adjusting the SLA threshold of the application; and

dynamically re-adjusting the SLA threshold of the application, to optimize the performance metric for the application.

8. The method as in claim 1 , wherein computing the CDF for the tracked performance metric comprises:

using a machine learning model to predict values of the performance metric for the tunnel; and

computing the CDF, at least in part, using the predicted values of the performance metric.

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

updating the CDF using additional values of the tracked performance metric for the tunnel.

10. The method as in claim 1 , wherein the tracked performance metric is indicative of: loss on the tunnel, latency on the tunnel, or jitter on the tunnel.

11. An apparatus, comprising:

one or more network interfaces to communicate with a software defined wide area network (SD-WAN);

a processor coupled to the network interfaces and configured to execute one or more processes; and

a memory configured to store a process executable by the processor, the process when executed configured to:

track a performance metric for a tunnel in the SD-WAN;

compute a cumulative distribution function (CDF) for the tracked performance metric;

assess curvature of the CDF for the tracked performance metric relative to a service level agreement (SLA) threshold of an application for that performance metric; and

control assignment of traffic for the application to the tunnel, based on the assessed curvature of the CDF for the tracked performance metric relative to the SLA threshold of the application for that performance metric,

wherein the apparatus assesses the curvature of the CDF for the tracked performance metric relative to the SLA threshold of the application for that performance metric by:

computing a point of slope increase on the CDF; and

determining a distance between the computed point of slope increase on the CDF and the SLA threshold of the application for that performance metric, wherein the computed point of slope increase on the CDF is either above or below the SLA threshold.

12. The apparatus as in claim 11 , wherein the apparatus controls assignment of the traffic for the application to the tunnel, based on the assessed curvature of the CDF for the tracked performance metric relative to the SLA threshold of the application for that performance metric, by:

determining that the SLA threshold is below the point of slope increase on the CDF beyond a predefined distance; and

assigning the traffic for the application to the tunnel, based on the determination.

13. The apparatus as in claim 11 , wherein the apparatus controls assignment of the traffic for the application to the tunnel, based on the assessed curvature of the CDF for the tracked performance metric relative to the SLA threshold of the application for that performance metric, by:

determining that the SLA threshold is above the point of slope increase on the CDF beyond a predefined distance; and

prohibiting the traffic for the application from being assigned to the tunnel, based on the determination.

14. The apparatus as in claim 11 , wherein the apparatus controls assignment of the traffic for the application to the tunnel, based on the assessed curvature of the CDF for the tracked performance metric relative to the SLA threshold of the application for that performance metric, by:

determining that the SLA threshold is near the point of slope increase on the CDF when the distance between the point of slope increase on the CDF and the SLA threshold is less than a predefined distance; and

flagging the tunnel as SLA threshold sensitive, based on the determination.

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

adjust the SLA threshold of the application, in an effort to make the tunnel non-SLA threshold sensitive; and

assign the traffic for the application to the tunnel, with the adjusted SLA threshold.

16. The apparatus as in claim 15 , wherein the process when executed is further configured to:

monitor a performance metric for the application, after adjusting the SLA threshold of the application; and

dynamically re-adjust the SLA threshold of the application, when the performance metric for the application is unacceptable.

17. The apparatus as in claim 16 , wherein the process when executed is further configured to:

monitor a performance metric for the application, after adjusting the SLA threshold of the application; and

dynamically re-adjust the SLA threshold of the application, to optimize the performance metric for the application.

18. The apparatus as in claim 11 , wherein the apparatus computes the CDF for the tracked performance metric by:

using a machine learning model to predict values of the performance metric for the tunnel; and

computing the CDF, at least in part, using the predicted values of the performance metric.

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

update the CDF using additional values of the tracked performance metric for the tunnel.

20. A tangible, non-transitory, computer-readable medium storing program instructions that cause a supervisory service for a software defined wide area network (SD-WAN) to execute a process comprising:

tracking, by the supervisory service for the SD-WAN, a performance metric for a tunnel in the SD-WAN;

computing, by the supervisory service, a cumulative distribution function (CDF) for the tracked performance metric;

assessing, by the supervisory service, curvature of the CDF for the tracked performance metric relative to a service level agreement (SLA) threshold of an application for that performance metric; and

controlling, by the supervisory service, assignment of traffic for the application to the tunnel, based on the assessed curvature of the CDF for the tracked performance metric relative to the SLA threshold of the application for that performance metric

wherein the assessing of the curvature of the CDF for the tracked performance metric relative to the SLA threshold of the application for that performance metric comprises:

computing, by the supervisory service, a point of slope increase on the CDF; and

determining, by the supervisory service, a distance between the computed point of slope increase on the CDF and the SLA threshold of the application for that performance metric, wherein the computed point of slope increase on the CDF is either above or below the SLA threshold.