Network Fault Impact Analysis Using Traffic Pattern Matching
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for identifying affected customers in data centers during communication faults are inefficient, as they rely on static configuration information and cannot accurately distinguish between affected and unaffected customers, especially in cloud services where real-time analysis of vast communication data is challenging.
Innovation Solution
A management apparatus that analyzes periodic communication patterns and calculates the certainty of customer impact by comparing fault information with historical traffic data, using a processor to determine the likelihood of customer involvement in network flows and outputting results for swift customer differentiation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If static configuration information is used to identify potentially affected customers, then the identification process is simple and fast, but it cannot accurately distinguish between affected and unaffected customers
Solution Approach 1:
The system performs preliminary analysis to determine communication patterns and relationships between customers and network components before faults occur. This pre-established knowledge base enables accurate identification of affected customers when faults happen, without requiring complex real-time analysis during incident response.
Solution Approach 2:
The system creates a virtual representation (copy) of the network topology and customer communication relationships based on observed traffic patterns. This virtual model can be queried rapidly during fault incidents to identify affected customers, avoiding the need for complex real-time analysis of actual network traffic.
2Productivity
If real-time analysis of all communications is performed to determine customer impact, then accurate identification of affected customers is possible, but it is difficult to analyze in real time due to vast amounts of communication data
Solution Approach 1:
The system extracts only the essential features from communication data - specifically, the relationships between customers and network components, and communication patterns over time. By extracting these key features rather than analyzing all raw communication data, the system achieves fast fault response without requiring complex real-time processing of vast data volumes.
Solution Approach 2:
The system performs preliminary analysis to build a knowledge base of communication patterns and customer-network relationships before faults occur. This pre-computed information enables rapid identification of affected customers during incidents, achieving high productivity without complex real-time analysis.
3Measurement precision
If packet sampling method is used to estimate network topology, then topology information can be obtained, but only communications with large traffic are sampled making it impossible to ascertain communication usage for users with small traffic
Solution Approach 1:
The system applies different analysis approaches to different data characteristics. For high-traffic communications, it uses efficient sampling methods, while for low-traffic communications, it employs more sensitive detection techniques. This localized adaptation of analysis methods ensures accurate detection of all communication usage regardless of traffic volume, without requiring exhaustive analysis of all data.
Data Source
AI summary
A management apparatus executes: receiving from a given communication apparatus where a fault has occurred, fault information; selecting a given flow that passes through the given communication apparatus from among a group of flows on the basis of the identification information of the given communication apparatus included in the fault information; determining whether the communication pattern of the given flow is similar to the given time series data indicating the change over time of the traffic within a period from a set period prior to the fault occurrence date and time to the fault occurrence date and time for the given flow; calculating the certainty that the given flow has been transmitted on the fault occurrence date and time on the basis of a frequency of appearance of the traffic of the communication pattern determined to be similar; and outputting calculation results.