Generic Network Analysis Model Decoupling Topology from Root Cause Diagnosis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional network management approaches face challenges in diagnosing and maintaining complex, distributed, and heterogeneous network infrastructures due to their reliance on topology-based analysis models, which are time-consuming and error-prone, especially in large networks with diverse domains like TCP/IP, SAN, and optical networks.
Innovation Solution
A generic analysis model based on causal relationships between network entities is introduced, decoupling topology data from analysis data, allowing for root cause and impact analysis independently of the network domain, and employing a codebook methodology to correlate symptoms with causes, thereby simplifying the analysis process and improving scalability.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional topology-based analysis models are used for network management, then network entities and their interconnections can be represented, but diagnosing and maintaining complex distributed networks becomes time-consuming and error-prone
Solution Approach 1:
The patent introduces an intermediary layer that transforms topology views into analysis views. This intermediary transformation process converts complex topological data into simplified analysis representations that enable faster and more accurate root cause diagnosis without losing essential network relationship information.
Solution Approach 2:
The patent segments the network analysis process into distinct views: topology views for representing network structure and analysis views for diagnosing issues. This segmentation allows the system to handle complex topological data separately from the diagnostic analysis, improving both accuracy and speed by focusing each view on its specific function.
2Adaptability or versatility
If topology views are used to represent network infrastructure, then network entities and interconnections are captured, but the analysis process becomes complex and domain-specific
Solution Approach 1:
The patent creates a universal analysis view that can be applied across multiple network domains (TCP/IP, SAN, optical networks). This analysis view serves as a domain-independent representation that simplifies the analysis process while maintaining applicability to various network types, eliminating the need for separate complex analysis models for each domain.
Solution Approach 2:
The transformation from topology views to analysis views acts as an intermediary process that eliminates domain-specific complexities. This intermediary layer translates diverse domain-specific topological representations into a unified analysis view that simplifies the diagnostic process while preserving the essential network relationships needed for cross-domain applicability.
3Measurement precision
If conventional codebook approaches are used for root cause analysis, then symptoms can be correlated with causes, but the codebook grows exponentially with network size making it non-scalable
Solution Approach 1:
The patent segments the codebook approach by creating domain-independent analysis views that separate the correlation logic from the network-specific data. This segmentation allows the same analysis framework to scale across networks of different sizes and domains without requiring exponential growth in codebook complexity, as the analysis view abstracts away the combinatorial explosion of possible symptom-cause relationships.
Data Source
AI summary
A generic analysis model is applicable to a variety of network infrastructure domains, and operable for causal analysis in a common manner independently of the corresponding underlying domain. The generic analysis model may be employed to model root cause analysis and impact analysis for different resource management systems such as enterprise networks, storage area networks, service provider networks and business process management. Such a model improves performance and scalability by simplifying analysis model for a given solution, and decouples the development from topology building and analysis implementation.