Resource Grouping for Performance Analysis in Large-Scale Systems
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In large-scale information systems, conventional performance monitoring methods struggle to analyze performance data due to complex interdependencies among resources, making it difficult to detect failures and abnormal behaviors efficiently.
Innovation Solution
A performance analysis method that divides resources into groups based on correlation of performance data changes, allowing for targeted analysis and modeling of each resource group to extract characteristic behaviors and detect anomalies.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional performance monitoring software is used to monitor all resources in a large-scale information system, then the number of output items becomes enormous, but it becomes difficult to perform analysis due to complex interdependencies
Solution Approach 1:
The patent segments the information system into multiple information systems based on correlation of performance data. By dividing the large-scale system into smaller, correlated subsystems, the analysis complexity is reduced while maintaining comprehensive monitoring coverage. Each segmented system can be analyzed independently, making the overall analysis manageable.
Solution Approach 2:
The patent introduces an intermediary process that calculates correlation between performance data of different resources and uses this correlation to automatically segment the system. This intermediary correlation analysis acts as a mediator between raw performance data and final analysis results, organizing the complex data into manageable groups.
2Reliability
If all performance data from numerous resources is analyzed together, then comprehensive monitoring is achieved, but detection of failures and abnormal behaviors becomes inefficient
Solution Approach 1:
By segmenting resources into correlated groups, the patent enables focused analysis on each segment rather than analyzing all resources simultaneously. This segmentation maintains reliable failure detection within each segment while significantly improving analysis efficiency by reducing the scope of analysis required.
Solution Approach 2:
The patent applies local quality by treating each segmented information system with analysis methods tailored to its specific characteristics and correlation patterns. Instead of applying a uniform analysis approach to all resources, each segment receives localized analysis appropriate to its nature, improving both detection reliability and efficiency.
3Measurement precision
If threshold-based monitoring is used for each output item, then abnormal behaviors can be detected, but the enormous number of output items makes it difficult to identify root causes
Solution Approach 1:
The patent segments the system so that threshold-based monitoring is applied to segmented groups rather than individual resources in isolation. This segmentation preserves anomaly detection accuracy within each segment while reducing the number of threshold checks needed, thereby preventing information loss about root causes.
Solution Approach 2:
The correlation-based segmentation process acts as an intermediary that organizes resources into meaningful groups before threshold monitoring is applied. This intermediary structure preserves the ability to detect anomalies accurately while maintaining cause identification capability by grouping related resources that share common failure modes.
Data Source
AI summary
A performance analysis method of a computer system using a management computer. The management computer includes: a processor; and a memory device in which a program to be executed by the processor is stored. The computer system is constituted by a plurality of resources. The processor divides the plurality of resources into a plurality of resource groups based on a correlation of changes in performance data between the resources, and analyzes the performance data for each of the divided resource groups.