Automated Remediation Confidence Control Through Network Path Observation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional automated remediation systems lack intelligence to learn from mistakes, fail to prevent repeat failures, and do not drive improvement in decision making or infrastructure stability without human input, leading to undesired impacts on critical services.
Innovation Solution
A closed-loop system that measures the success of auto-remediation tasks using historical confidence scores and service health measurements, adjusting deployment of automated responses based on complex rules and human reinforcement to ensure intelligent feedback and improved decision making.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If automated remediation actions are deployed without feedback mechanisms, then automation speed and response time are improved, but reliability and service stability deteriorate due to inability to learn from mistakes
Solution Approach 1:
The patent implements feedback mechanisms where service health measurements and path change observations are collected after automated remediation actions are executed. This feedback is used to update confidence scores associated with automated actions, enabling the system to learn from past successes and failures. The feedback loop ensures that automated actions are continuously improved based on actual service impact, resolving the contradiction between fast automated response and reliable remediation.
2Extent of automation
If automated remediation actions are executed without monitoring service impact, then automation extent is improved, but harmful effects on critical services increase due to chain reactions
Solution Approach 1:
The patent performs service health measurements and path change observations before and after automated remediation actions to predict and prevent harmful chain reactions. By monitoring service impact in advance and observing path changes, the system can identify potential negative effects before they manifest as service degradation. This cushioning approach allows extensive automation while protecting against harmful side effects through proactive monitoring and analysis.
3Device complexity
If conventional automated actions are deployed without learning mechanisms, then system simplicity is maintained, but adaptability to prevent repeat failures is reduced
Solution Approach 1:
The patent implements self-service learning mechanisms where the automated remediation system automatically collects service health data, analyzes path changes, updates confidence scores, and improves its own decision-making without external human intervention. The system serves itself by learning from executed actions and using this knowledge to prevent repeat failures, achieving adaptability while maintaining relatively simple system architecture through automated self-improvement processes.
Data Source
AI summary
A method comprises: at a controller of a network, establishing a historical probability of success that an automated response, when executed on the network, will remediate an alarm event: performing a service test on the network to collect first path information from the network; after performing the service test, receiving from the network a report of the alarm event, and deploying the automated response to the network to remediate the alarm event; after the automated response is executed, repeating the service test on the network to collect second path information and a second test result that indicates pass or fail; determining whether a change from the first path information to the second path information represents a network degradation; and adjusting the historical probability of the success based on the second test result and whether the change represents the network degradation.


