PIN Blades for Disaggregated Data Center Network Congestion

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

The increasing number of devices interconnected in disaggregated data centers leads to network congestion and delay fluctuations, particularly at top-of-rack and spine switches.

Innovation Solution

The introduction of Processing-in-Network (PIN) blades, which include packet reception and transmission units, functional units for data processing, and flow ID switches to manage and route data between resource blades, reducing network load and congestion by performing data processing and transmission tasks within the network.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Adaptability or versatility

If resources are disaggregated into separate resource blades interconnected by a network, then resource flexibility and independent advancement are improved, but network congestion and delay fluctuation increase

Engineering Contradiction:

Improveresource flexibilityVSAvoidnetwork congestion

Core Design Contradiction:

Adaptability or versatilityVSReliability

Solution Approach 1:

The patent introduces processing blocks as intermediary components between resource blades and network switches. These processing blocks perform packet processing and data forwarding functions locally, reducing the burden on central switches and eliminating network congestion while maintaining the flexibility of disaggregated resource architecture

Inventive Principle:

Principle #24Intermediary (Mediator)

Solution Approach 2:

The patent segments the network processing function by distributing processing blocks to each resource blade. This segmentation allows local packet processing and data forwarding, reducing centralized switch congestion while maintaining network flexibility and adaptability

Inventive Principle:

Principle #1Segmentation

2Productivity

If the number of interconnected devices increases in disaggregated data centers, then resource scalability is improved, but network congestion and delay fluctuation worsen

Engineering Contradiction:

Improveresource scalabilityVSAvoidcommunication delay

Core Design Contradiction:

ProductivityVSLoss of time

Solution Approach 1:

The patent implements preliminary packet processing and data forwarding at the processing blocks located at resource blades. By performing packet processing locally before data reaches the network switch, the system reduces transmission time and eliminates delay fluctuation while supporting scalable resource configurations

Inventive Principle:

Principle #10Preliminary action

3Speed

If data processing is performed at resource blades, then processing speed is improved, but device complexity increases

Engineering Contradiction:

Improveprocessing speedVSAvoidprocessing block complexity

Core Design Contradiction:

SpeedVSDevice complexity

Solution Approach 1:

The patent designs processing blocks with multi-functional capabilities that can perform various packet processing operations (routing, switching, data forwarding) depending on the resource blade type. This universality allows a single processing block design to serve multiple functions across different resource blades, increasing processing speed without proportionally increasing device complexity

Inventive Principle:

Principle #6Universality (Multi-functionality)

Data Source

PatentUS20240414103A1Computing system

Publication Date: 2024.12.12 NIPPON TELEGRAPH & TELEPHONE CORP

AI summary

A computing system includes top-of-rack switches, a spine switch, and PIN blades arranged between the top-of-rack switches and resource blades. Each of the PIN blades includes a plurality of processing blocks, and each of the processing blocks includes a functional unit that performs predetermined processing on data included in the received packet.