Middleware Performance Diagnosis via Trend Correlation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current approaches for diagnosing performance problems in application server environments are not standardized or automated across various containers, making it difficult to determine the root cause of issues effectively.
Innovation Solution
A method that analyzes performance metrics over different time periods, correlates trend information with a knowledge base, and alerts users to potential causes of abnormal behavior, using a combination of diagnostic agents, collectors, and a problem tree to identify bottlenecks and root causes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Extent of automation
If current diagnostic approaches are used, then administrators can identify performance problems, but the process cannot be standardized and automated across various containers
Solution Approach 1:
The patent creates a universal diagnostic framework that works across multiple container types (servlet containers, EJB containers, messaging containers, etc.) by defining a common problem tree structure and standardized performance indicators that can be applied uniformly across different container technologies, enabling both automation and standardization simultaneously
2Measurement precision
If administrators manually analyze performance metrics, then they can identify root causes, but the process requires complete knowledge of all components and vast information
Solution Approach 1:
The patent introduces an intermediary diagnostic system that includes a problem tree structure and correlation engine, which mediates between raw performance metrics and root cause identification. This intermediary layer processes and correlates multiple performance indicators automatically, eliminating the need for administrators to have complete knowledge of all system components while maintaining high diagnostic accuracy
3Reliability
If performance metrics are analyzed over a longer time period, then trend information is obtained, but the response time to identify problems increases
Solution Approach 1:
The patent performs preliminary analysis by continuously collecting and pre-processing performance indicators in the background, maintaining a problem tree structure that is partially built in advance. When anomalies occur, the system can quickly correlate against pre-established patterns and relationships, reducing the time needed to identify root causes while still utilizing extended time period data for accurate trend analysis
Data Source
AI summary
A method of determining a root cause of a performance problem is provided. The method comprises analyzing a plurality of performance indicators/metrics in a first time period and determining that at least one performance indicators/metric is exhibiting abnormal behavior. The method further comprises analyzing the plurality of performance indicators/metrics over a second time period, the second time period is longer than the first time period, and determining trend information for each performance indicators/metric over the second time period. The method further comprises correlating the trend information for each performance indicators/metric with performance problem information stored in a knowledge base, identifying a potential cause of the abnormal behavior based on the correlation, and alerting a user of the potential cause.