SD-WAN Traffic Engineering Using Predicted Metrics to Reduce Bottlenecks

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

SD-WAN networks face challenges in optimizing traffic engineering due to dynamic network conditions, leading to bottlenecks and suboptimal user experiences despite existing traffic steering mechanisms.

Innovation Solution

Implementing a prediction engine at a centralized controller to generate predicted performance metrics using machine learning and AI, which are then distributed to network devices for making proactive traffic engineering decisions based on future network conditions.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If traditional traffic engineering based on current network conditions is used, then the system is simple to operate, but it cannot anticipate future bottlenecks and provides suboptimal user experience

Engineering Contradiction:

Improveuser experience qualityVSAvoidsystem complexity

Core Design Contradiction:

ReliabilityVSDevice complexity

Solution Approach 1:

The system performs preliminary actions by using machine learning models to predict future network performance metrics before bottlenecks occur. The controller generates predicted performance metrics based on historical and current data, enabling proactive traffic engineering decisions that prevent future network issues rather than merely reacting to current conditions.

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

An intermediary prediction engine is introduced between the network devices and the traffic engineering controller. This intermediary component processes network data through machine learning algorithms to generate predicted performance metrics, which then guide traffic engineering decisions, bridging the gap between raw network data and actionable insights.

Inventive Principle:

Principle #24Intermediary (Mediator)

2Reliability

If real-time traffic steering based on current metrics is implemented, then the response time is fast, but it cannot prevent future bottlenecks

Engineering Contradiction:

Improvenetwork performanceVSAvoidreactive delay

Core Design Contradiction:

ReliabilityVSLoss of time

Solution Approach 1:

The system performs traffic engineering actions in advance by predicting future network conditions. Instead of waiting for bottlenecks to manifest in real-time metrics, the prediction engine identifies potential issues beforehand and triggers preventive traffic steering actions, eliminating reactive delays and improving overall network performance.

Inventive Principle:

Principle #10Preliminary action

3Reliability

If predictive analytics are added to the network system, then future network conditions can be anticipated, but the computational requirements and system complexity increase

Engineering Contradiction:

Improveprediction accuracyVSAvoidcomputational resources

Core Design Contradiction:

ReliabilityVSUse of energy by moving object

Solution Approach 1:

The system segments the predictive analytics functionality into a dedicated prediction engine that operates separately from core network forwarding functions. This segmentation allows ML model training and prediction operations to be performed on specialized hardware or cloud resources, reducing the computational burden on network devices themselves while maintaining high prediction accuracy.

Inventive Principle:

Principle #1Segmentation

Data Source

PatentUS20250219946A1Methods and systems for prediction based traffic engineering at a network device

Publication Date: 2025.07.03 VERSA NETWORKS

AI summary

Methods and system for operating an network device are disclosed. In an embodiment, a method for operating a network device involves generating a traffic engineering decision in response to applying a traffic engineering rule to a predicted performance metric, and implementing a traffic engineering action at the network device in response to the traffic engineering decision.