Drift Detection via Incident Parameters in Node Meshes
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Detecting the causes of drift in resource usage patterns across interconnected computing resources is challenging, especially when multiple infrastructure components are involved, leading to difficulties in identifying issues or anticipating component failures.
Innovation Solution
The approach involves creating a mesh representation of the topology of physical and/or logical nodes, using incident parameters to determine the value of drift due to incidents, and calculating a drift value using empirically known incident probability values to differentiate between genuine incidents and other changes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If multiple infrastructure components are monitored for resource usage patterns, then the ability to detect drift improves, but the complexity of determining the cause of changes increases
Solution Approach 1:
The patent segments the complex monitoring system into distinct modules: resource usage pattern detection, drift determination, incident parameter identification, and cause classification. Each module handles a specific aspect of the analysis, breaking down the complex task of determining drift causes into manageable segments that can be processed independently and systematically.
Solution Approach 2:
The patent introduces incident parameters as intermediary elements that mediate between resource usage changes and their causes. These incident parameters serve as intermediate representations that capture the relationship between drift events and underlying incidents, making the cause determination process more tractable by providing structured intermediate data for analysis.
2Measurement precision
If incident parameters are used to determine drift values, then the accuracy of drift detection improves, but the computational resources required increase
Solution Approach 1:
The patent performs preliminary actions by pre-defining incident parameters and their relationships with resource usage patterns. By establishing the framework of incident parameters and their expected impacts on resource usage in advance, the system avoids performing complex computations for every drift event, instead leveraging pre-established knowledge to efficiently determine drift values when incidents occur.
3Reliability
If drift detection is performed across interconnected computing resources, then the ability to identify issues improves, but the time required for analysis increases
Solution Approach 1:
The patent segments the analysis across different levels: individual resource usage patterns, drift events, incident parameters, and cause categories. This hierarchical segmentation allows parallel processing at each level, reducing overall analysis time while maintaining comprehensive issue identification across interconnected computing resources.
Solution Approach 2:
The patent implements feedback mechanisms where detected drift events trigger incident parameter analysis, which in turn provides feedback for classifying causes. This feedback loop enables rapid iteration and refinement of issue identification, reducing analysis time by focusing computational efforts on the most likely causes based on observed drift patterns and incident parameter relationships.
Data Source
AI summary
In certain embodiments, a computer-implemented method includes: discovering, by a drift detection device executing on one or more processors, a mesh of nodes topology; generating a plurality of subgraphs using the mesh of nodes topology, wherein each of the plurality of subgraphs comprises a portion of a plurality of nodes of the mesh of nodes; selecting a subgraph of the plurality of subgraphs; calculating a Drift-subgraph (Dsg) value for the subgraph using a plurality of subgraph incident parameter values; and calculating a drift value (D) using the Dsg value and an Incident likelihood (IL) value, wherein the drift value indicates whether drift occurred in the subgraph.