Deep Packet Buffering for Ethernet Switch Bandwidth Fluctuations

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Ethernet switch cores face challenges in managing momentary bandwidth variances, which can lead to packet loss, especially in environments with significant fluctuations due to weather or air interfaces, where standard flow control is insufficient.

Innovation Solution

The implementation of a deep packet buffering system in Ethernet switch cores, utilizing both on-chip and external deep packet buffers, with policy engines to determine frame destinations and manage bandwidth variance, transferring frames to external buffers only when necessary to maintain lossless operation.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If standard on-chip buffering is used in Ethernet switch cores, then device complexity is kept low and operation speed is maintained, but packet loss occurs during momentary bandwidth reductions

Engineering Contradiction:
Improvepacket loss preventionVSAvoidbuffering system complexity
Core Design Contradiction:
ReliabilityVSDevice complexity

Solution Approach 1:

The buffering system is segmented into multiple levels: on-chip buffers for immediate packet storage and external deep packet buffers for extended storage during bandwidth fluctuations. This segmentation allows the system to maintain low complexity for normal operation while providing enhanced reliability during bandwidth reductions without requiring a completely complex buffering architecture.

Inventive Principle:
Principle #1Segmentation

2Reliability

If deep packet buffering is implemented to absorb bandwidth fluctuations, then packet loss is reduced, but device complexity and resource requirements increase

Engineering Contradiction:
Improvelossless operationVSAvoidbuffering infrastructure complexity
Core Design Contradiction:
ReliabilityVSDevice complexity

Solution Approach 1:

The buffering architecture uses a nested structure where external deep packet buffers are coupled with internal on-chip buffers. The on-chip buffers serve as a first level of buffering integrated within the switch core, while external buffers provide a second level for deeper buffering needs. This nesting allows the system to achieve lossless operation by transferring packets between buffer levels based on bandwidth conditions, without requiring a completely separate complex buffering system.

Inventive Principle:
Principle #7Nested doll (Nesting)

3Reliability

If momentary bandwidth reductions are absorbed by the backhaul network infrastructure, then packet loss is prevented, but the system requires significant buffering capacity that increases device complexity

Engineering Contradiction:
Improvepacket delivery reliabilityVSAvoidbuffering capacity
Core Design Contradiction:
ReliabilityVSQuantity of substance

Solution Approach 1:

The buffering system operates dynamically by monitoring bandwidth conditions and adaptively transferring packets between on-chip and external buffers. During momentary bandwidth reductions, the system activates deep packet buffering in external memory to absorb the fluctuations. During normal operation, packets remain in the faster on-chip buffers. This dynamic operation allows the system to provide significant buffering capacity only when needed, reducing the average buffering requirements and associated device complexity.

Inventive Principle:
Principle #15Dynamics

Data Source

PatentUS8855127B2Method and system for intelligent deep packet buffering
Publication Date: 2014.10.07 AVAGO TECHNOLOGIES INTERNATIONAL SALES PTE LTD
  • US8855127B2 patent drawing
  • US8855127B2 patent drawing
  • US8855127B2 patent drawing

AI summary

Disclosed is a method and system for deep packet buffering on a switch core comprising an ingress and egress deep packet buffer and an external deep packet buffer.