Cross-Layer Node Scoring for Root Cause Alert Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Identifying the root cause of failures in an information processing system is challenging due to the complex interdependencies among nodes in different layers, making it difficult to pinpoint the node responsible for alerts generated by abnormalities.
Innovation Solution
An information processing apparatus calculates base scores based on caller-callee relationships within layers and propagation scores based on dependency relationships across layers, using PageRank and propagation score coefficients to determine failure scores for each node, thereby identifying nodes likely to be the cause of alerts.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional monitoring methods sum points for each server based on CPU/memory usage and process counts, then the system can identify failure causes, but the accuracy is insufficient due to complex interdependencies among nodes in different layers
Solution Approach 1:
The patent segments the information processing system into multiple layers (infrastructure layer, virtualization layer, operating system layer, application layer) and calculates failure scores independently for each node within these layers. By dividing the complex system into manageable layers and using PageRank to assess individual node importance within each layer, the method achieves accurate failure cause identification without being overwhelmed by overall system complexity
Solution Approach 2:
The patent introduces a new dimension of analysis by calculating propagation scores that measure the influence of adjacent nodes across different layers. Instead of only analyzing nodes within the same layer, the method evaluates vertical dependencies between layers (e.g., how an infrastructure node affects application nodes), adding a cross-layer dimensional perspective that improves identification accuracy while systematically handling complexity
2Measurement precision
If the system calculates failure scores considering all node interdependencies, then the identification accuracy improves, but the computational complexity increases
Solution Approach 1:
The patent divides the computational task into two separate scoring mechanisms: PageRank for assessing intrinsic node importance within layers, and propagation score for evaluating cross-layer influences. This segmentation allows the system to handle complex interdependencies through modular calculations rather than attempting to compute all dependencies simultaneously, reducing computational power requirements while maintaining assessment accuracy
Solution Approach 2:
The patent focuses computational resources on calculating scores for nodes adjacent to failed nodes and their immediate dependencies, rather than uniformly processing all nodes in the system. By applying propagation scores selectively to relevant adjacent nodes and using the failure score formula only for nodes near failures, the method achieves accurate identification without the excessive computational cost of analyzing the entire system
3Measurement precision
If the system uses PageRank to calculate base scores based on caller-callee relationships, then the node importance within layers is accurately reflected, but the cross-layer dependencies are not sufficiently considered
Solution Approach 1:
The patent segments the scoring function into two distinct components: PageRank for intra-layer importance assessment and propagation score for inter-layer dependency evaluation. The base score from PageRank captures caller-callee relationships within the same layer, while the separate propagation score mechanism specifically addresses cross-layer influences, ensuring both intra-layer accuracy and inter-layer adaptability through divided functional responsibility
Solution Approach 2:
The patent merges the base score (intra-layer importance) and propagation score (inter-layer influence) into a comprehensive failure score using the formula: failure score = base score + propagation score. This combination integrates both intra-layer PageRank accuracy and inter-layer dependency coverage, allowing the system to simultaneously reflect node importance within layers and their influences across layers in a unified assessment
4Measurement precision
If the system calculates propagation scores for all node pairs, then the cross-layer influence is fully captured, but the processing time increases
Solution Approach 1:
The patent applies propagation score calculations locally only to nodes adjacent to failed nodes and their immediate dependencies, rather than uniformly calculating scores for all node pairs in the system. This localized approach captures the essential cross-layer influences on failed nodes while avoiding the time-consuming computation of propagation scores for distant or irrelevant nodes, significantly reducing processing time
Data Source
AI summary
A processing unit calculates a base score coefficient indicating a weight for each node, based on caller-callee relationship between nodes in the same layer as indicated in configuration information, calculates a base score for each node, based on alert information and the base score coefficients, the base score being based on an alert, calculates, for each pair of a node and its adjacent node identified based on the configuration information, a propagation score by multiplying the base score of the adjacent node by a propagation score coefficient based on dependency relationship between the node and the adjacent node, calculates, for each node, a failure score by summing the base score of the node and the propagation score corresponding to a pair of the node and its adjacent node, and identifies one or more nodes that are candidates for the cause of the alert, based on the failure scores.