Spine-and-leaf network switch with fail-in-place redundancy

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current network switch architectures face limitations in packaging due to quadratic growth in externally facing ports, leading to economic constraints and increased complexity as port speeds and numbers increase, particularly in high-speed and large-port-count configurations.

Innovation Solution

The proposed solution involves a spine-and-leaf array architecture where spine cards are surrounded by leaf cards on multiple sides, exploiting redundancy to allow spine ports or chips to fail without replacement, redirecting traffic to working ports, thereby increasing the number of externally facing ports significantly while maintaining high throughput.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Quantity of substance

If traditional network switch architecture is used, then packaging complexity increases quadratically with port count, but this leads to economic constraints and increased complexity as port speeds and numbers increase

Engineering Contradiction:
Improvenumber of externally facing portsVSAvoidpackaging complexity
Core Design Contradiction:
Quantity of substanceVSDevice complexity

Solution Approach 1:

The switch is divided into modular spine cards and leaf cards that can be independently packaged and configured. Each spine card contains multiple spine chips that can be separately replaced if failing, and each leaf card contains leaf chips connected to external ports. This segmentation allows linear scalability of port count without quadratic packaging complexity increase.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent transitions from a traditional two-dimensional front-side packaging approach to a three-dimensional configuration where leaf cards surround spine cards on multiple sides. This spatial reconfiguration enables significantly more externally facing ports to be accommodated within the same footprint, reducing packaging density constraints.

Inventive Principle:
Principle #17Another dimension (Dimensionality change)

2Ease of repair

If spine cards are made accessible for replacement, then ease of repair improves, but this precludes the fail-in-place strategy and reduces redundancy benefits

Engineering Contradiction:
Improvespine card replaceabilityVSAvoidfail-in-place capability
Core Design Contradiction:
Ease of repairVSReliability

Solution Approach 1:

The spine card is segmented into multiple independently replaceable spine chips. When a failure occurs, only the specific failing chip needs to be replaced rather than the entire spine card, maintaining fail-in-place capability while enabling targeted repair. This segmentation preserves both reliability and ease of repair.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

Multiple spine chips are implemented as identical copies on each spine card, allowing any single chip failure to be masked by the remaining functional chips. This redundancy enables fail-in-place operation where the system continues to function with reduced capacity rather than requiring complete spine card replacement.

Inventive Principle:
Principle #26Copying

3Quantity of substance

If more SERDES channels are used to increase port count, then externally facing ports increase, but power consumption and costs increase

Engineering Contradiction:
Improvenumber of externally facing portsVSAvoidpower consumption
Core Design Contradiction:
Quantity of substanceVSUse of energy by moving object

Solution Approach 1:

The spine chips serve multiple functions: they provide internal interconnection between leaf cards and also serve as externally facing ports when leaf cards are removed or fail. This multi-functionality allows the same hardware infrastructure to support more ports without proportionally increasing SERDES channel count, thereby reducing power consumption.

Inventive Principle:
Principle #6Universality (Multi-functionality)

Solution Approach 2:

The system implements more spine chips than strictly necessary for minimal connectivity, creating excess capacity that can be utilized when leaf cards fail or are removed. This partial over-provisioning allows the system to maintain functionality with fewer active SERDES channels, reducing overall power consumption while supporting high port counts.

Inventive Principle:
Principle #16Partial or excessive action

Data Source

PatentUS10389654B2High-density, fail-in-place switches for computer and data networks
Publication Date: 2019.08.20 INTERNATIONAL BUSINESS MACHINE CORPORATION
  • US10389654B2 patent drawing
  • US10389654B2 patent drawing
  • US10389654B2 patent drawing

AI summary

A structure for a network switch. The network switch may include a plurality of spine chips arranged on a plurality of spine cards, where one or more spine chips are located on each spine card; and a plurality of leaf chips arranged on a plurality of leaf cards, wherein one or more leaf chips are located on each leaf card, where each spine card is connected to every leaf chip and the plurality of spine chips are surrounded on at least two sides by leaf cards.