Network Congestion Module for Microburst Detection

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current network elements struggle to accurately determine the cause and effect of congestion in real-time due to large polling periods, which can miss microbursts and fail to provide timely information on when congestion occurs, what causes it, and what is affected by it.

Innovation Solution

Implementing a network congestion module that measures queue group occupancy in near-real-time using ASICs, with different thresholds to monitor and record queue group occupancy, gather statistics, and perform actions to characterize packets and alleviate congestion, allowing for precise detection of congestion causes and effects.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Device complexity

If the network element uses a large polling period to check dropped packet counters, then the device complexity is reduced and energy consumption is lowered, but the measurement precision of congestion detection deteriorates and microbursts are missed

Engineering Contradiction:
Improvecongestion detection mechanismVSAvoidcongestion detection precision
Core Design Contradiction:
Device complexityVSMeasurement precision

Solution Approach 1:

The patent segments the congestion detection function into two parts: (1) continuous queue occupancy measurement by hardware counters that track queue depth at high resolution, and (2) periodic analysis of dropped packet counters. This segmentation allows precise real-time detection without requiring the entire system to operate at high complexity continuously.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent introduces queue occupancy counters as an intermediary measurement mechanism. Instead of directly polling dropped packet counters at small intervals, the system uses queue occupancy counters that continuously track queue depth, providing precise congestion indicators without requiring frequent polling of the main control system.

Inventive Principle:
Principle #24Intermediary (Mediator)

2Measurement precision

If the network element polls dropped packet counters frequently with a small polling period, then the measurement precision of congestion detection is improved and microbursts are detected, but the device complexity increases and processing overhead rises

Engineering Contradiction:
Improvecongestion detection precisionVSAvoidcongestion detection mechanism
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent implements preliminary action by having hardware counters continuously measure and record queue occupancy depth in real-time before congestion occurs. This pre-measurement approach eliminates the need for frequent polling during congestion events, as the data is already captured and ready for analysis.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The queue occupancy counters operate autonomously in the data plane, self-measuring queue depth without requiring control plane intervention. This self-service mechanism provides continuous precise measurements while keeping the control plane simple and reducing processing overhead.

Inventive Principle:
Principle #25Self-service

3Productivity

If the network element uses a large polling period, then the loss of time for congestion detection is reduced in terms of processing overhead, but the actual detection time lag increases making it difficult to determine cause and effect

Engineering Contradiction:
Improvepolling efficiencyVSAvoidcongestion detection time lag
Core Design Contradiction:
ProductivityVSLoss of time

Solution Approach 1:

The patent ensures continuity of useful action by having queue occupancy counters continuously measure queue depth at all times, not just during periodic polling. This continuous measurement eliminates detection time lags while maintaining high polling efficiency, as the data is always available when needed for analysis.

Inventive Principle:
Principle #20Continuity of useful action

4Loss of information

If the network element implements detailed packet statistics gathering and analysis, then the information completeness about congestion cause and effect is improved, but the processing overhead and device complexity increase

Engineering Contradiction:
Improvecongestion information completenessVSAvoidpacket analysis mechanism
Core Design Contradiction:
Loss of informationVSDevice complexity

Solution Approach 1:

The patent applies partial action by gathering packet statistics selectively based on queue occupancy thresholds. Instead of analyzing all packets continuously, the system captures detailed packet information only when queue depth exceeds predefined thresholds, providing sufficient congestion information while minimizing processing overhead.

Inventive Principle:
Principle #16Partial or excessive action

Data Source

PatentEP2779542B1System and method for determining a cause of network congestion
Publication Date: 2020.04.22 ARISTA NETWORKS INC
  • EP2779542B1 patent drawingFigure 1
  • EP2779542B1 patent drawingFigure 2
  • EP2779542B1 patent drawingFigure 3

AI summary

A method and apparatus of a device that determines a cause and effect of congestion in this device is described. In an exemplary embodiment, the device measures a queue group occupancy of a queue group for a port in the device, where the queue group stores a plurality of packets to be communicated through that port. In addition, the device determines if the measurement indicates a potential congestion of the queue group, where the congestion prevents a packet from being communicated within a time period. If potential congestion exists on that queue group, the device further gathers information regarding packets to be transmitted through that port. For example, the device can gather statistics packets that are stored in the queue group and/or new enqueue packets.