Error Cause Determination via Dependency Graph Suspect Scores
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Determining the cause of an error state in computing environments is laborious, time-consuming, and error-prone due to the need for manual analysis of log files and correlation of past events, especially when multiple elements impact a single element's state.
Innovation Solution
A dependency graph is created to represent the computing environment, where a first node represents the impacted element and impacting nodes represent elements that can cause an error state, with a suspect score computed for each impacting element based on the number of elements it affects and those in an error state, allowing for accurate identification of the cause without requiring the impacting element to be in an error state.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If manual analysis of log files and correlation of past events is performed to determine the cause of error state, then measurement precision is improved, but loss of time increases and productivity decreases
Solution Approach 1:
The system pre-establishes a dependency graph representing relationships between computing elements before errors occur. This graph is built in advance using known dependencies (e.g., power supply to server, network switch to computer), so when an error occurs, the system can immediately query pre-defined relationships rather than manually analyzing log files and correlating events in real-time.
Solution Approach 2:
The patent introduces an intermediary system that computes suspect scores for impacting elements based on the dependency graph and current error states. This intermediary computation layer automatically correlates events and identifies probable causes without requiring manual log analysis, thereby reducing time loss while maintaining measurement precision through systematic scoring.
2Measurement precision
If manual analysis of log files and correlation of past events is performed to determine the cause of error state, then measurement precision is improved, but productivity decreases
Solution Approach 1:
The dependency graph is constructed in advance, encoding all known relationships between computing elements. This preliminary structuring of knowledge allows the system to rapidly identify probable causes when errors occur, eliminating the need for manual log analysis and event correlation, thereby significantly improving productivity in error state restoration while maintaining accurate cause determination.
Solution Approach 2:
The system performs automatic computation of suspect scores and identification of probable error causes without requiring manual intervention. The intermediary system self-services by querying the dependency graph, analyzing current error states, and computing probabilities automatically, thus improving productivity while maintaining measurement precision through systematic automated analysis.
3Device complexity
If the impacting element must be in an error state to be identified as the cause, then measurement precision is simplified, but measurement precision decreases because impacting elements may not change their own state
Solution Approach 1:
The patent introduces an intermediary computation system that calculates suspect scores based on the dependency graph and the error states of impacted elements. This intermediary layer enables the system to identify probable causes even when impacting elements are not in error states themselves, by inferring causality from the relationships encoded in the dependency graph and the observed error states of dependent elements.
Solution Approach 2:
The system changes the parameter used for cause identification from 'error state of impacting element' to 'suspect score computed from dependency relationships and impacted element states'. This parameter transformation allows accurate identification of causes even when the impacting element does not exhibit an error state, resolving the contradiction between simplicity and accuracy.
Data Source
AI summary
In an example, suspect scores for impacting elements that can cause an error state of a first element are computed. The computation is performed based on states of each element for which the impacting element can cause the error state.