Dynamic Threshold Anomaly Detection for Web Performance Monitoring
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing web performance monitoring systems face challenges in accurately detecting and analyzing anomalies due to the distributed nature of web-based resources, as they often generate false alarms from fluctuating network traffic and cannot account for peak and trough conditions, leading to inefficient detection and analysis of web performance anomalies.
Innovation Solution
A method and system for analyzing time-series data that dynamically adapts performance metric monitoring by using a stream-processing apparatus to aggregate data, a statistical hypothesis test to detect deviations from baseline values, and a call graph representation to identify root causes of anomalies, allowing for real-time or near-real-time analysis and alert generation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If a fixed threshold is set for performance metrics, then the monitoring system is simple to operate, but it generates false alarms during peak network traffic and misses anomalies during trough periods
Solution Approach 1:
The patent applies dynamics by replacing fixed thresholds with dynamically adapting thresholds that automatically adjust based on network traffic conditions. The system monitors performance metrics continuously and updates thresholds in response to changing network conditions, allowing the monitoring system to maintain high detection accuracy without manual intervention while adapting to peak and trough periods automatically
Solution Approach 2:
The patent changes the parameter of threshold values from static to dynamic based on network traffic conditions. By adjusting threshold parameters in response to observed network behavior patterns, the system eliminates false alarms during peak traffic while maintaining sensitivity during normal periods, resolving the contradiction between operational simplicity and detection reliability
2Measurement precision
If custom thresholds are set for each data center, then detection precision for each location is improved, but the system complexity increases and cannot account for network traffic fluctuations
Solution Approach 1:
The patent applies self-service by enabling the system to automatically adjust its own detection parameters without external intervention. Each data center's performance metrics are monitored and used to automatically update local thresholds, allowing the system to maintain high detection precision for each location while eliminating the need for manual threshold configuration and reducing overall system complexity
Solution Approach 2:
The patent implements feedback mechanisms where performance metric data from each data center is continuously fed back to update the thresholds. This closed-loop system automatically adapts thresholds based on actual network conditions at each location, maintaining precision without requiring complex manual configuration or external intervention
3Adaptability or versatility
If manual threshold selection is used to accommodate different data center locations, then the system can handle distributed architecture, but it cannot dynamically respond to changing network conditions
Solution Approach 1:
The patent applies dynamics by transitioning from static manual threshold selection to dynamic automatic threshold adjustment. The system continues to support multiple data center locations through distributed monitoring but automatically adapts thresholds in real-time based on local network conditions, enabling rapid response to changing conditions without manual reconfiguration at each location
Data Source
AI summary
A system for processing data is provided. During operation, the system obtains a current window of one or more intervals of timeseries data collected from a monitored system. Next, the system continuously performs a statistical hypothesis test that compares the one or more intervals of the time-series data with baseline values from historic time-series data associated with the monitored system. When the statistical hypothesis test indicates a deviation of the time-series data from the baseline values, the system outputs an alert of an anomaly represented by the deviation.