Ethernet Gateway Packet Delivery to Memory Fabric Queues

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing network architectures face inefficiencies in processing ingress packets, as they often require servers to handle and process packets, which can lead to resource consumption and reduced scalability, especially in high-performance memory fabric environments.

Innovation Solution

The Ethernet gateway parses ingress packets to determine if they can be directly delivered to a destination queue at an egress Ethernet gateway, bypassing servers and improving throughput by redirecting ingress traffic directly to the memory fabric, thereby reducing server processor load and enhancing cost efficiency.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If servers process ingress packets through traditional network architecture, then packet processing can be performed with existing infrastructure, but server processor load increases and scalability decreases

Engineering Contradiction:
Improvepacket throughputVSAvoidserver processing load
Core Design Contradiction:
ProductivityVSDevice complexity

Solution Approach 1:

The patent extracts the packet processing function from servers and relocates it to network switches. The switch directly delivers ingress packets to egress queues without server intervention, removing the processing burden from servers while maintaining packet forwarding capability through the network infrastructure.

Inventive Principle:
Principle #2Taking out (Extraction)

Solution Approach 2:

The network switch acts as an intermediary between ingress and egress packets. It receives ingress packets, determines the appropriate egress queue based on destination information, and delivers packets directly to the correct queue without requiring server processing, thus mediating the packet flow efficiently.

Inventive Principle:
Principle #24Intermediary (Mediator)

2Reliability

If ingress packets are delivered through servers to queues, then packet routing can be managed centrally, but resource consumption increases and cost efficiency decreases

Engineering Contradiction:
Improvepacket delivery accuracyVSAvoidresource consumption
Core Design Contradiction:
ReliabilityVSUse of energy by moving object

Solution Approach 1:

The network switch performs self-service by autonomously determining egress queue destinations and delivering packets directly without server involvement. The switch uses its own routing logic and queue management capabilities to handle packet delivery, eliminating the need for server resources while maintaining accurate packet routing.

Inventive Principle:
Principle #25Self-service

3Adaptability or versatility

If traditional server-based packet processing is used, then existing network architecture can be maintained, but scalability and throughput are limited

Engineering Contradiction:
Improvenetwork scalabilityVSAvoidnetwork throughput
Core Design Contradiction:
Adaptability or versatilityVSProductivity

Solution Approach 1:

The patent segments the packet processing function from the server system and assigns it to the network switch. This segmentation allows the server to focus on higher-level processing while the switch handles packet forwarding, enabling independent optimization and scaling of each component without being constrained by server processing capacity.

Inventive Principle:
Principle #1Segmentation

Data Source

PatentEP3491792B1Deliver an ingress packet to a queue at a gateway device
Publication Date: 2021.02.17 HEWLETT PACKARD ENTERPRISE DEV LP
  • EP3491792B1 patent drawingFigure 1
  • EP3491792B1 patent drawingFigure 2
  • EP3491792B1 patent drawingFigure 3

AI summary

The disclosure relates to a method, system, and medium to deliver an ingress packet to a queue at a gateway device. In an example, the gateway device can receive an ingress packet from an Ethernet. Moreover, the gateway device determines a memory queue identifier corresponding to a queue at a second gateway device for the ingress packet. Then, the gateway device delivers the ingress packet to the queue at the second gateway device without the ingress packet being processed at the plurality of server processors.