SD-WAN Path Selection Using Polling Metrics
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current software-defined wide area networks (SD-WANs) face challenges in selecting optimal paths for data traffic due to reliance on probe-based metrics, which may not accurately reflect real-time connectivity and can lead to suboptimal routing, especially in dynamic network conditions.
Innovation Solution
A method that involves polling peers for traffic metrics, combining and calculating performance across multiple paths using Dijkstra Algorithm with weight ranges, and selecting the path based on thresholds to ensure optimal latency and packet loss, with the ability to reroute in case of link failures.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If probe-based metrics are used for path selection, then path selection can be implemented, but the metrics do not accurately reflect real-time connectivity leading to suboptimal routing
Solution Approach 1:
The patent performs preliminary path calculations using Dijkstra's algorithm with pre-configured weight ranges and thresholds before actual data traffic needs routing. Traffic metrics are polled in advance from peers, and multiple candidate paths are pre-computed and stored, so when routing decisions are needed, the system can quickly select from pre-evaluated options rather than computing metrics in real-time, thus achieving both accuracy and efficiency
Solution Approach 2:
The patent implements dynamic path selection by continuously polling peers for updated traffic metrics and recalculating path performance using Dijkstra's algorithm with configurable weight ranges. The system adapts to changing network conditions by dynamically adjusting which paths are selected based on current metric values against predefined thresholds, ensuring routing decisions reflect real-time connectivity status
2Reliability
If multiple paths are calculated and evaluated, then optimal path selection is achieved, but the computational complexity increases
Solution Approach 1:
The patent simplifies the complexity of evaluating multiple paths by changing the parameters used in path calculation. Instead of evaluating all possible paths with equal detail, the system uses Dijkstra's algorithm with pre-configured weight ranges and thresholds that prune the search space. Paths are evaluated using modified cost functions that incorporate multiple traffic metrics weighted according to policy, reducing computational complexity while maintaining routing optimality
Solution Approach 2:
The patent segments the path selection process into distinct phases: (1) polling peers for traffic metrics, (2) combining metrics from multiple sources, (3) calculating path performance using Dijkstra's algorithm with weight ranges, (4) evaluating candidate paths against thresholds, and (5) selecting the optimal path. This segmentation allows each phase to be optimized independently and reduces overall computational complexity by breaking down the complex routing decision into manageable steps
Data Source
AI summary
Embodiments herein disclose path selection for data traffic within a software-defined wide area network using traffic metrics. Some embodiments relate to a method that includes polling peers of the SD-WAN for traffic metrics, receiving traffic metrics from at least a portion of the peers, combining the received traffic metrics, calculating performance of a plurality of possible paths from a source node to a destination node, the nodes being within the SD-WAN topology, and selecting a path from the source node to the destination node based on the calculated performance.