Ontology-Based Root Cause Analysis for Business Application Services
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for diagnosing faults in business network nodes are insufficient as they only consider direct cause-and-effect relationships, leading to inappropriate remedial actions when indirect relationships are involved, causing the root cause of failures to be overlooked.
Innovation Solution
The system generates an ontology of business application nodes and their relationships, using data collection agents to monitor connections and state changes, allowing for the identification of both direct and indirect root causes of errors, enabling accurate fault diagnosis.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional direct cause-and-effect monitoring is used, then the monitoring method is simple, but the root cause detection accuracy deteriorates when indirect relationships are involved
Solution Approach 1:
The patent segments the monitoring system into multiple intelligence levels: edge intelligence at individual nodes for local anomaly detection, cluster intelligence for intermediate analysis, and cloud intelligence for comprehensive root cause determination. This segmentation allows the system to achieve high detection accuracy through distributed analysis while managing complexity through hierarchical organization of monitoring functions.
Solution Approach 2:
The patent introduces cluster intelligence as an intermediary layer between edge nodes and cloud infrastructure. This intermediary performs intermediate analysis of anomalies detected at edge nodes, filtering and preprocessing data before sending to the cloud, thereby improving root cause detection accuracy while reducing the computational burden and complexity at any single level.
2Measurement precision
If comprehensive node relationship analysis is implemented, then fault diagnosis accuracy improves, but the analysis time increases
Solution Approach 1:
The patent implements preliminary action by continuously building and maintaining an ontology model of node relationships, traffic patterns, and dependencies in advance. When faults occur, the system can immediately query this pre-established model rather than analyzing relationships from scratch, thereby achieving high fault diagnosis accuracy while minimizing analysis time.
Solution Approach 2:
The patent applies dynamics by implementing real-time adaptation of the ontology model based on changing network conditions, traffic patterns, and node states. The system dynamically updates relationship weights and anomaly thresholds, allowing comprehensive analysis to remain accurate while adapting to current system state, thus reducing the time needed for accurate fault diagnosis.
Data Source
AI summary
An ontology is generated for a business application on an enterprise network that describes one or more nodes that communicate with each other during the execution of the business application. An alert condition of the business application is detected, and the ontology for the business application is processed to determine one or more components of the ontology that are in an alert state. Further, a root cause view that indicates the one or more alert state components is generated and displayed to a user.