Automated Outlier Detection for Computer System Maintenance
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Identifying and resolving system defects and optimizations in complex computer systems is a tedious, time-consuming, and expensive process, especially as the number of systems increases, requiring manual combing through vast amounts of data to detect issues in mission-critical applications.
Innovation Solution
A system and method where multiple computer systems communicate data to a central system that uses data normalization algorithms to analyze and identify outlier systems, comparing them to representative expectations to detect unexpected performance, and distributing optimization or defect solutions back to the reporting systems.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If manual methods are used to comb through system data to identify defects and optimizations, then detailed analysis can be performed, but the process becomes tedious, time-consuming, and expensive as the number of systems increases
Solution Approach 1:
The patent replaces manual mechanical analysis with automated computer-based analysis. The system uses automated data collection, normalization, and comparison processes to identify outlier systems, substituting human effort with computational algorithms that can process vast amounts of system data quickly and accurately without the time constraints of manual review
Solution Approach 2:
The system enables self-service by automatically collecting data from multiple systems, performing normalization and comparison operations, and identifying defects without requiring human intervention. The automated outlier detection process allows the system to serve itself by continuously monitoring and analyzing its own performance data across the fleet
2Reliability
If manual analysis is used to ensure systems are optimally configured and defect-free, then system reliability can be maintained, but the process becomes increasingly complex and resource-intensive as the number of systems scales
Solution Approach 1:
The patent transforms the complexity of analyzing multiple systems by changing the parameters of analysis through data normalization. By normalizing diverse system data to common standards and parameters, the system can compare apples to apples across different hardware configurations, operating systems, and workloads, maintaining reliability without proportionally increasing analysis complexity
Solution Approach 2:
The system segments the analysis process into distinct automated components: data collection from multiple systems, normalization of data to standard parameters, comparison against representative models, and identification of outliers. This segmentation allows each component to be handled independently by automated processes, preventing complexity from compounding as system numbers increase
3Loss of information
If comprehensive data collection from multiple systems is performed to identify patterns and defects, then better system understanding is achieved, but the amount of data to be processed increases significantly
Solution Approach 1:
The system extracts only the essential and comparable parameters from comprehensive system data through normalization. Rather than processing all raw data from multiple systems, the process extracts key performance metrics and configuration parameters that can be meaningfully compared, discarding redundant or non-comparable information while retaining the critical data needed for defect identification
Solution Approach 2:
The patent adds the dimension of normalization to the data processing approach. By transforming diverse system parameters into a normalized dimensional space with common units and scales, the system can process comprehensive data from multiple systems without being overwhelmed by heterogeneity, enabling pattern recognition across the entire data set
Data Source
AI summary
Intrinsic system metadata is received from the plurality of computer systems, wherein the intrinsic system metadata includes operational data reported by individual systems in the plurality of computer systems. The intrinsic system metadata is normalized to identify a representative. Individual outlier systems are identified by comparing operational data to the representative.