Network Processor Queue Control for External Memory Bandwidth Collapse

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Network processors face bandwidth collapse due to external memory overload from rogue source devices or microbursts, as external memory bandwidth cannot keep pace with line bandwidth, leading to throughput limitations and resource waste.

Innovation Solution

Implementing a bandwidth utilization mechanism and queue size reduction in network processors to monitor and manage external memory usage, sending congestion notifications and automatically resizing bad queues to prevent overload.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Ease of manufacture

If external memory bandwidth is reduced to lower costs, then device cost decreases, but external memory becomes prone to overload and bandwidth collapse occurs

Engineering Contradiction:
Improvedevice costVSAvoidexternal memory reliability
Core Design Contradiction:
Ease of manufactureVSReliability

Solution Approach 1:

The patent applies preliminary action by proactively reducing the queue size of identified bad queues before external memory becomes overloaded. The system continuously monitors queue depths and packet rates, and when a queue is identified as bad (unresponsive to congestion notifications), its maximum depth is reduced in advance. This preventive approach stops harmful traffic patterns before they can overwhelm the limited external memory bandwidth, thereby maintaining reliability while keeping external memory bandwidth low for cost reasons.

Inventive Principle:
Principle #10Preliminary action

2Adaptability or versatility

If deep queues are used to perform AQM functions, then congestion management capability improves, but external memory bandwidth becomes the bottleneck limiting line throughput

Engineering Contradiction:
Improvecongestion management capabilityVSAvoidline throughput
Core Design Contradiction:
Adaptability or versatilityVSProductivity

Solution Approach 1:

The patent applies local quality by differentiating between good queues and bad queues, and applying different queue depth configurations to each. Good queues (those that respond to congestion notifications) maintain deep queue depths to enable effective AQM functions. Bad queues (those that do not respond) are assigned reduced queue depths to prevent them from consuming excessive external memory bandwidth. This localized differentiation allows the system to maintain congestion management capability for legitimate traffic while preventing bandwidth collapse from misbehaving sources.

Inventive Principle:
Principle #3Local quality

3Productivity

If queue size is reduced for bad queues, then external memory bandwidth consumption decreases, but congestion management precision may be affected

Engineering Contradiction:
Improveexternal memory bandwidth utilizationVSAvoidcongestion management precision
Core Design Contradiction:
ProductivityVSMeasurement precision

Solution Approach 1:

The patent applies feedback by continuously monitoring queue depths, packet rates, and congestion notification responses to dynamically identify bad queues. The system uses this feedback information to make intelligent decisions about which queues to shrink and which to maintain at full depth. This feedback mechanism ensures that congestion management precision is preserved for good queues while allowing aggressive bandwidth control for bad queues, thereby achieving both goals simultaneously.

Inventive Principle:
Principle #23Feedback

Data Source

PatentEP4195616B1Network processor with external memory protection
Publication Date: 2026.03.04 NOKIA SOLUTIONS & NETWORKS OY
  • EP4195616B1 patent drawingFigure 1
  • EP4195616B1 patent drawingFigure 2
  • EP4195616B1 patent drawingFigure 3

AI summary

Systems and methods for protecting external memory resources to prevent bandwidth collapse in a network processor. One embodiment is a network processor including an input port configured to receive packets from a source device, on-chip memory configured to store packets in queues, an external memory interface configured to couple the on-chip memory with an external memory providing a backing store to the on-chip memory, and bandwidth monitor configured to measure a bandwidth utilization of the external memory. The network processor also includes a processor configured to apply the bandwidth utilization of the external memory to a congestion notification profile, to generate one or more congestion notifications based on the bandwidth utilization applied to the congestion notification profile, and to send the one or more congestion notifications to the source device to request decreasing packet rate for decreasing the bandwidth utilization of the external memory.