Network Path Health Scoring for Transient Latency Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing network path monitoring systems fail to effectively detect transient or ephemeral issues that impact application performance and violate service level agreements due to the dynamic nature of network loads and route changes in software-defined wide area networks (SD-WANs).
Innovation Solution
A network path scoring system that calculates a nearly real-time (NRT) health score based on current latency, a non-stationary range of expected/acceptable additional latency, bandwidth capacity, and load, using a formula that adjusts to load variations and route changes, providing intuitive scores on a 0 to 100 scale.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional network monitoring systems are used, then network path health can be monitored, but transient or ephemeral issues impacting application performance cannot be effectively detected
Solution Approach 1:
The system dynamically adjusts the baseline latency and latency range based on historical data and current network conditions. Instead of using fixed thresholds, the monitoring adapts to changing network characteristics, enabling detection of transient issues that deviate from the dynamic baseline while maintaining reliability under varying loads and routes.
Solution Approach 2:
The system pre-calculates expected latency ranges and baselines before actual performance evaluation occurs. By establishing what normal performance should be under various conditions in advance, the system can quickly identify deviations indicating transient issues without waiting for problems to manifest, thus improving detection precision while ensuring service level agreement compliance.
2Adaptability or versatility
If network monitoring adapts to load variations and route changes, then detection accuracy improves, but system complexity increases
Solution Approach 1:
The system changes key parameters (baseline latency, latency range, scoring thresholds) based on observed network conditions rather than using complex adaptive algorithms. By adjusting these parameters dynamically based on simple metrics like historical latency data and current load, the system achieves adaptability to load variations and route changes while maintaining relatively simple system architecture.
Solution Approach 2:
The monitoring system automatically adjusts its own parameters and baselines without requiring manual configuration or complex external control. It self-adapts to network changes by continuously learning from historical data and automatically recalibrating its expectations, reducing the need for complex manual management while maintaining high adaptability to network dynamics.
3Productivity
If nearly real-time scoring is implemented, then timely detection of network problems is enabled, but computational requirements increase
Solution Approach 1:
The system performs scoring at selective intervals rather than continuously, and focuses computational effort on calculating only the necessary metrics (current latency, baseline comparison, score assignment). By performing partial calculations at strategically chosen moments rather than exhaustive continuous analysis, the system achieves timely problem detection while reducing overall computational resource consumption.
Solution Approach 2:
The system uses lightweight, easily computable metrics (simple latency measurements, basic statistical comparisons) rather than complex long-term analyses. Each scoring event is a discrete, inexpensive calculation that can be quickly discarded and replaced with the next measurement, enabling frequent near-real-time scoring without excessive computational burden.
Data Source
AI summary
A network path scoring system is disclosed herein that presents scores of the “health” of network paths graphically based on additional latencies of the network path across a plurality of time intervals. The system scores health of a network path based on additional latency of the network path for a current time interval, additional latency expected for the network path, and the current load. The scoring uses a non-stationary range that is based on expected additional latency (e.g., engineered/injected latency) and coefficient(s) that vary with load. The graphically presented health scores of the path are then updated with the health score of the current time interval in real-time or near real-time.