Network Event Flood Prediction via Dynamic Rate Trending
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing network management systems face challenges in managing large networks during cascade failures, where a flood of fault events can overwhelm the system, leading to data loss and unresponsive management, and existing predictive analytics rely on fixed thresholds, failing to adapt to device-specific fault rates effectively.
Innovation Solution
An apparatus and method that predict network event floods by detecting event rates, aggregating and trending data, generating maximum acceptable event rate levels, and signaling a predicted flood, allowing for intelligent flood protection mode activation and deactivation of devices to manage data flow effectively.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If a monitoring probe initiates shutdown once fault event rate exceeds a threshold, then the system protects itself from event flood, but data loss occurs and cascade failure may already have started
Solution Approach 1:
The system performs preliminary action by predicting event floods before they occur using statistical analysis of event rate trends. The predictor compares current event rates against historically-derived thresholds to anticipate future floods, enabling proactive mitigation rather than reactive shutdown. This allows the system to prepare flood protection measures in advance while maintaining normal operation and data collection.
Solution Approach 2:
The system implements feedback by continuously monitoring event rates, comparing them against dynamic thresholds, and adjusting flood protection measures based on the comparison results. The threshold itself is derived from historical feedback about system behavior patterns. This closed-loop approach enables the system to respond appropriately to changing conditions while minimizing data loss through intelligent, data-driven decisions.
2Adaptability or versatility
If existing predictive analytics use fixed thresholds, then the system can detect abnormal event rates, but it cannot adapt to device-specific fault rates effectively
Solution Approach 1:
The system applies local quality by deriving device-specific event rate thresholds from individual device historical data rather than using uniform system-wide thresholds. Each device develops its own baseline event rate characteristics, allowing the predictive analytics to adapt to device-specific fault patterns. This localized approach maintains effectiveness across diverse device types without requiring complex device-specific configurations.
Solution Approach 2:
The system implements self-service by automatically learning device-specific event rate patterns from historical data without requiring manual configuration or expert intervention. The predictive analytics engine autonomously derives thresholds for each device based on its own operational history, enabling effective device-specific adaptation while keeping the system relatively simple to deploy and maintain.
3Productivity
If a probe monitors multiple devices, then comprehensive monitoring is achieved, but all data from all devices is lost when flood protection activates
Solution Approach 1:
The system applies segmentation by implementing device-specific flood protection controls that can be applied independently to individual devices or groups of devices. Rather than treating all monitored devices as a single unit, the system can selectively activate flood protection for specific devices experiencing abnormal event rates while continuing to monitor and collect data from other devices normally. This granular approach preserves comprehensive monitoring coverage while minimizing data loss.
Solution Approach 2:
The system implements dynamics by making flood protection status dynamic and adjustable at the device level rather than static and uniform across all devices. The flood protection state can change independently for each device based on its own event rate characteristics and predictions. This dynamic, device-level control allows the system to maintain comprehensive monitoring while selectively protecting against floods only where necessary.
Data Source
AI summary
An apparatus for predicting a network event flood comprises an event rate detector for detecting rates of event emissions from one or more devices; an aggregator for producing an aggregate rate and an aggregate rate trend of the rates of event emissions from a plurality of the devices; a level generator for generating a plurality of levels comprising maximum acceptable event rate values of a plurality of the aggregate rate trends over plural time periods; a storage component for storing the plurality of levels; a comparator for comparing a current aggregate rate trend with at least a selected one of the levels; and a signaller for signalling a predicted event flood responsive to the comparator detecting that the current aggregate rate trend will exceed the at least a selected one of the levels at a first point in time.