Network Observability for Root Cause Detection Under Traffic Rerouting
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The complexity of global computer networks makes it difficult to track and monitor the health of each network, leading to challenges in identifying the root cause of performance anomalies, as traditional methods often fail to pinpoint the exact location of issues, especially when networks reroute traffic around the problematic area.
Innovation Solution
A device employs machine learning-based network observability process that analyzes time series of network metrics, identifies a loss of visibility, and determines the root cause of performance anomalies by correlating end-to-end and hop-by-hop probing results, providing an indication of the specific location responsible for the anomaly.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional network monitoring methods are used, then network operators can monitor network health, but they cannot accurately identify the root cause of performance anomalies when networks reroute traffic
Solution Approach 1:
The patent introduces an intermediary observability system that sits between the network traffic and the monitoring point. This system captures and analyzes telemetry data from multiple sources (end-to-end probes, hop-by-hop probes, network metrics) to reconstruct the network state even when direct traffic flow is rerouted. The intermediary system correlates data from different probing methods to identify the root cause location without requiring direct traffic passage through the problematic network segment.
Solution Approach 2:
The patent segments the network monitoring approach into multiple independent probing methods: end-to-end probing, hop-by-hop probing, and collection of network metrics from various sources. Each segmentation targets a specific aspect of network behavior. By combining results from these segmented approaches, the system can triangulate the root cause location even when individual methods are obscured by traffic rerouting.
2Adaptability or versatility
If network operators use educated guesses to identify problematic networks, then they can make hypotheses about issues, but they lack sufficient data to support or reject their hypotheses
Solution Approach 1:
The patent implements a feedback mechanism where the observability system continuously collects network telemetry data, compares it against expected baseline behavior, and provides feedback to confirm or refute network operator hypotheses. The system generates actionable insights that directly test hypotheses by providing empirical evidence from multiple probing methods, allowing operators to validate or reject their educated guesses with concrete data.
Solution Approach 2:
The patent creates a universal observability platform that performs multiple functions: collecting network metrics, executing end-to-end probes, performing hop-by-hop probing, analyzing telemetry data, and generating insights for hypothesis validation. This multi-functional system replaces multiple specialized tools with a single platform that can address various network monitoring needs and validate different types of hypotheses.
3Adaptability or versatility
If the network complexity increases with more interconnected networks and geographically dispersed resources, then network coverage and service availability improve, but the difficulty of tracking and monitoring network health increases
Solution Approach 1:
The patent merges multiple monitoring approaches (end-to-end probing, hop-by-hop probing, network metrics collection) into a single unified observability system. This consolidation reduces the complexity of managing separate monitoring tools while maintaining comprehensive network coverage across interconnected networks and geographically dispersed resources. The merged system processes and correlates data from all sources through a single analysis engine.
Data Source
AI summary
In one embodiment, a device detects a performance anomaly in a network based on a plurality of time series of network metrics collected from the network. The device determines a loss of visibility in the network based on results from probing the network. The device identifies a particular location in the network as a root cause of the performance anomaly based on the loss of visibility. The device provides, to a user interface, an indication of the particular location being the root cause of the performance anomaly.