Automated Error Analysis for Storage System Test Logs
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for reviewing test execution logs in storage systems are tedious and time-consuming, requiring extensive manual labor to identify and correlate errors, especially in complex testing environments where hundreds or thousands of errors are reported weekly.
Innovation Solution
An automated error analysis system using unsupervised learning for clustering and supervised learning processes, specifically density-based spatial clustering of applications with noise and random forest deep neural networks, to quickly identify error locations and types in test execution logs, reducing manual effort and enhancing accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If manual review methods are used to analyze test execution logs, then error identification can be performed with simple tools, but the process becomes extremely time-consuming and labor-intensive when hundreds or thousands of errors are reported weekly
Solution Approach 1:
The patent replaces manual mechanical review processes with automated machine learning systems. Specifically, unsupervised learning algorithms (clustering, anomaly detection) and supervised learning algorithms (classification models) automatically analyze test execution logs, identify error patterns, and correlate errors with failure events, eliminating the need for manual log review and dramatically increasing analysis throughput
Solution Approach 2:
The patent introduces machine learning models as intermediary systems between raw test execution logs and error identification results. These models process the logs, extract features, identify patterns, and generate error correlations, serving as an intelligent mediator that transforms unstructured log data into actionable error insights without human intervention
2Measurement precision
If extensive manual labor is applied to correlate errors in complex testing environments, then comprehensive error analysis can be achieved, but the complexity and resource requirements of the process increase significantly
Solution Approach 1:
The patent segments the error analysis process into distinct automated stages: log parsing and feature extraction, unsupervised learning for pattern identification (clustering, anomaly detection), supervised learning for classification, and error correlation generation. Each stage is handled by specialized algorithms, reducing overall system complexity while maintaining high correlation accuracy through systematic decomposition of the analysis task
Solution Approach 2:
The patent transforms unstructured log data into structured features through parameter extraction and normalization. By converting raw log entries into standardized numerical or categorical features, the system enables efficient machine learning processing and accurate error correlation without requiring complex manual analysis procedures
3Productivity
If automated machine learning processes are implemented to identify error patterns, then analysis speed and accuracy improve, but the system complexity and computational resources required increase
Solution Approach 1:
The patent implements a dynamic, two-phase learning system that adapts to different analysis needs. The first phase uses unsupervised learning (clustering, anomaly detection) to identify patterns without predefined labels, while the second phase applies supervised learning (classification models) for precise error categorization. This dynamic approach allows the system to handle both exploratory analysis and targeted classification, improving speed and accuracy while managing complexity through phased processing
Data Source
AI summary
A method of processing test execution logs to determine error location and source includes creating a set of training examples based on previously processed test execution logs, clustering the training examples into a set of clusters using an unsupervised learning process, and using training examples of each cluster to train a respective supervised learning process to label data where each generated cluster is used as a class/label to identify the type of errors in the test execution log. The labeled data is then processed by supervised learning processes, specifically a classification algorithm. Once the classification model is built it is used to predict the type of the errors in future/unseen test execution logs. In some embodiments, the unsupervised learning process is a density-based spatial clustering of applications with noise clustering application, and the supervised learning processes are random forest deep neural networks.