Adaptive SD-WAN Tunnel Stress Testing for SLA Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional SD-WAN failure detection is reactive, leading to traffic disruption until rerouting occurs, and training models for what-if scenarios in SD-WANs is challenging due to complexity and system dynamics, making simulations difficult.
Innovation Solution
A service computes expected information gain to reroute traffic from a primary tunnel to a backup tunnel, obtaining performance measurements to train a machine learning model predicting whether the rerouting satisfies the service level agreement (SLA), using adaptive stress testing to generate training data for what-if scenario model training.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If reactive failure detection using keep-alive mechanisms is used, then tunnel failures can be detected and traffic rerouted, but traffic disruption occurs before rerouting can take place
Solution Approach 1:
The system performs preliminary actions by proactively testing backup tunnels before primary tunnel failures occur. Stress tests are initiated on backup tunnels to pre-validate their capacity and performance, so when a failure occurs, the backup is already verified and ready for immediate traffic switching, eliminating traffic disruption time
Solution Approach 2:
The system uses its own traffic to perform self-testing of backup tunnels. By diverting a portion of its own traffic through backup paths during normal operation, the system automatically validates backup tunnel performance without external intervention, enabling continuous readiness verification
2Reliability
If simulations are used to train machine learning models for what-if scenarios, then predictive failure detection becomes possible, but the complexity and system dynamics of SD-WANs make simulations difficult to implement
Solution Approach 1:
Instead of creating complex simulations of the entire SD-WAN system, the system creates simplified copies or representations of tunnel behavior through actual stress testing. Real traffic patterns and performance metrics from actual tunnel stress tests provide authentic training data that captures system dynamics without requiring complex simulation models
Solution Approach 2:
The system changes parameters dynamically during stress testing by varying traffic load, tunnel conditions, and routing scenarios. This generates diverse training data across different operational states, enabling the machine learning model to learn predictive patterns from real parameter variations rather than static simulations
3Reliability
If traffic is rerouted onto a backup tunnel, then continuity of service is maintained, but the backup tunnel may not meet the service level agreement (SLA) of the rerouted traffic
Solution Approach 1:
The system performs preliminary validation of backup tunnel capacity through stress testing before relying on them for traffic rerouting. By pre-measuring performance metrics such as bandwidth, latency, and packet loss under various load conditions, the system ensures backup tunnels meet SLA requirements before they are needed, guaranteeing both continuity and quality
Solution Approach 2:
The system implements feedback mechanisms that continuously monitor backup tunnel performance and compare it against SLA thresholds. Performance measurements from stress tests and ongoing monitoring provide feedback that validates whether backup tunnels maintain acceptable service levels, enabling informed routing decisions
Data Source
AI summary
In one embodiment, a service in a network computes an expected information gain associated with rerouting traffic from a first tunnel onto a backup tunnel in the network. The service initiates, based on the expected information gain, rerouting of the traffic from the first tunnel onto the backup tunnel. The service obtains performance measurements for the traffic rerouted onto the backup tunnel. The service uses the performance measurements to train a machine learning model to predict whether rerouting traffic from the first tunnel onto the backup tunnel will satisfy a service level agreement (SLA) of the traffic.