Tiled Compute Node On-Chip Switching for Scalable Server Networking

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Traditional server infrastructure lacks flexibility in scaling networking components as more processing nodes are added, leading to challenges in load balancing and other datacenter tasks beyond computation requirements.

Innovation Solution

A server compute node with multiple tiled compute nodes, each equipped with a processor and on-chip layer 2 switch, enables scalable node-stacking and integrated networking functions, allowing nodes to connect directly and provide a single network interface without external networking components, using software and hardware to manage load balancing and routing.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If traditional server infrastructure is scaled by adding more processing nodes, then processing power is improved, but networking component flexibility and infrastructure complexity worsen

Engineering Contradiction:

Improveprocessing powerVSAvoidnetworking infrastructure complexity

Core Design Contradiction:

ProductivityVSDevice complexity

Solution Approach 1:

The patent combines networking functions (switching, load balancing, NAT, firewall) directly into the processing nodes themselves. Each processing node integrates both compute capabilities and networking services, eliminating the need for separate external networking hardware. This merging allows the system to scale processing power while maintaining networking flexibility, as each node independently provides full networking functionality.

Inventive Principle:

Principle #5Merging (Combining)

2Reliability

If external networking components are added to support more processing nodes, then network connectivity is improved, but infrastructure cost and device complexity increase

Engineering Contradiction:

Improvenetwork connectivityVSAvoidexternal networking components

Core Design Contradiction:

ReliabilityVSDevice complexity

Solution Approach 1:

Each processing node is designed to be self-sufficient with integrated networking capabilities. The nodes provide their own switching, load balancing, and security functions without requiring external networking infrastructure. This self-service approach maintains reliable network connectivity while reducing overall system complexity and eliminating the need for additional external networking components as the system scales.

Inventive Principle:

Principle #25Self-service

3Productivity

If processing nodes are added to increase computation capacity, then productivity is improved, but load balancing and networking flexibility worsen

Engineering Contradiction:

Improvecomputation capacityVSAvoidload balancing flexibility

Core Design Contradiction:

ProductivityVSAdaptability or versatility

Solution Approach 1:

The processing nodes are designed with universal networking functionality that handles multiple roles simultaneously. Each node can perform computing tasks, switching, load balancing, NAT, and firewall services. This multi-functionality allows the system to increase computation capacity while maintaining load balancing flexibility, as each node can dynamically adapt to different networking roles and workloads without requiring specialized external components.

Inventive Principle:

Principle #6Universality (Multi-functionality)

Data Source

PatentUS10887238B2High performance, scalable multi chip interconnect

Publication Date: 2021.01.05 MELLANOX TECHNOLOGIES LTD(IL)

AI summary

A flexible, scalable server is described. The server includes plural server nodes each server node including processor cores and switching circuitry configured to couple the processor to a network among the cores with the plurality of cores implementing networking functions within the compute nodes wherein the plurality of cores networking capabilities allow the cores to connect to each other, and to offer a single interface to a network coupled to the server.