SDN QoS Monitoring via Time Series Similarity
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing network monitoring techniques face challenges in efficiently and accurately monitoring large-scale network flows, particularly in detecting Quality of Service (QoS) degradation without impacting the network and are not applicable to all network flows, especially brown-outs which degrade application performance without causing complete failures.
Innovation Solution
A method and system that configure network elements to report statistical information as time series data, compute the similarity of this data, and indicate QoS degradation when it falls below a specified threshold, allowing for scalable, end-to-end QoS monitoring and diagnosis in Software Defined Networks (SDNs) with minimal overhead.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If existing monitoring techniques monitor network infrastructure and devices, then service outages can be detected, but brown-outs and QoS degradation are difficult to detect
Solution Approach 1:
The patent introduces an intermediary component (monitoring element) that collects statistical information from multiple network elements and performs similarity computation to detect QoS degradation. This intermediary layer enables accurate detection of brown-outs without requiring complex instrumentation of each individual network device, thus improving measurement precision while managing device complexity.
Solution Approach 2:
The patent replaces traditional active probing mechanisms with passive statistical information collection from network elements. Instead of injecting test traffic to detect QoS issues, the system utilizes existing network flow statistics and computes similarity metrics, reducing the mechanical complexity of the monitoring system while improving detection accuracy for subtle QoS degradation.
2Measurement precision
If large scale flow monitoring is performed to detect brown-outs, then QoS degradation can be detected, but network overhead increases
Solution Approach 1:
The patent extracts only the essential statistical information needed for QoS monitoring from network elements, rather than collecting complete flow data. By taking out only the necessary metrics (packet counts, byte counts, timestamps) and computing similarity based on these extracted features, the system achieves accurate brown-out detection while minimizing network overhead and energy consumption.
Solution Approach 2:
The patent applies partial monitoring by selecting specific network elements along the flow path rather than monitoring all elements. The similarity computation uses a subset of statistical data points sufficient for detection purposes, avoiding the excessive action of complete end-to-end flow monitoring while maintaining adequate detection precision for QoS degradation.
3Measurement precision
If active probing is used to monitor network flows, then QoS metrics can be measured, but network performance is impacted
Solution Approach 1:
The patent implements self-service monitoring where network elements themselves generate and report the statistical information needed for QoS monitoring. The monitored network infrastructure provides the measurement data without external probing, eliminating the harmful impact of active probes on network performance while maintaining measurement precision through the self-collected statistics.
Solution Approach 2:
The patent substitutes mechanical active probing with a passive information collection mechanism. Instead of using external probes to measure QoS metrics, the system relies on statistical information naturally generated by network elements during normal operation, thereby eliminating the performance impact associated with active probing while maintaining measurement accuracy.
4Measurement precision
If application instrumentation is used to monitor QoS, then accurate flow monitoring is achieved, but applicability is limited to instrumented applications
Solution Approach 1:
The patent creates a universal monitoring solution that works across diverse applications and network flows without requiring application-specific instrumentation. By placing the monitoring logic in the network infrastructure itself (at network elements and the monitoring element), the system achieves multi-functionality and broad adaptability while maintaining accurate flow monitoring through standardized statistical collection and similarity-based analysis.
Data Source
AI summary
A method for detecting Quality of Service degradation in a network flow includes collecting time series data representing statistical information pertaining to a network flow registered with a network flow monitoring service. The time series data is collected from at least two network elements on a path of the network flow configured to report the time series data. The method further includes indicating Quality of Service degradation in the network flow based on a similarity of the time series data to expected time series data being below a specified similarity threshold, and triggering at least one action to address the similarity being below the specified similarity threshold.