Predictive Routing for SD-WAN Breaking Point Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current technologies in software-defined wide area networks (SD-WANs) fail to accurately identify application performance breaking points, relying on static thresholds and reactive routing, which leads to poor Quality of Experience (QoE) due to unpredictable network dynamics and varying application sensitivities to path metrics.
Innovation Solution
A predictive routing process that uses machine learning to model uncertainty in application experience metrics, identifying breaking points in path metrics where the application experience degrades, and actively suggests new threshold levels for testing, integrating with routing mechanisms to initiate corrective measures such as rerouting traffic to better paths.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If static thresholds and reactive routing are used, then device complexity is reduced, but measurement precision of application performance breaking points deteriorates
Solution Approach 1:
The system performs preliminary actions by proactively predicting application performance breaking points using machine learning models before actual performance degradation occurs. The ML model analyzes historical path metrics and application performance data to forecast breaking points, enabling preventive routing decisions rather than reactive responses to failures.
Solution Approach 2:
The system implements continuous feedback loops where actual application performance measurements are compared against predicted breaking points. This feedback mechanism refines the machine learning models over time, improving breaking point detection accuracy while maintaining automated routing decisions that adapt to changing application sensitivities.
2Measurement precision
If machine learning models are deployed to predict breaking points, then measurement precision improves, but device complexity increases
Solution Approach 1:
The predictive routing system is segmented into distinct functional modules: data collection components that gather path metrics, machine learning model components that predict breaking points, and routing decision components that act on predictions. This segmentation allows each module to be optimized independently and deployed flexibly across different network elements.
Solution Approach 2:
The system introduces machine learning models as intermediary components between raw network metrics and routing decisions. These ML intermediaries process and interpret complex path metric data, translating it into actionable breaking point predictions that guide routing behavior without requiring direct complex rule-based logic in the routing engine itself.
3Reliability
If application-specific breaking points are identified, then reliability of service level agreements improves, but loss of information increases due to varying application sensitivities
Solution Approach 1:
The system applies local quality by identifying and enforcing application-specific breaking points tailored to each application's unique sensitivity characteristics. Rather than using uniform thresholds, the ML model learns and applies customized breaking point criteria for different applications based on their individual performance requirements and tolerances.
Solution Approach 2:
The system dynamically adjusts routing parameters based on identified application-specific breaking points. When path metrics approach these customized thresholds, the routing mechanism changes parameters such as route selection, traffic engineering settings, or resource allocation to maintain performance within acceptable ranges for each specific application.
Data Source
AI summary
In one embodiment, a device obtains path metrics for a network path via which traffic for an online application is conveyed. The device models uncertainty of an application experience metric predicted for the online application based on the path metrics. The device identifies, based on the uncertainty of the application experience metric modeled by the device, a breaking point in the path metrics at which the application experience metric predicted for the online application is expected to switch from being acceptable to being degraded. The device provides the breaking point in the path metrics for display.