Distributed Gateway Packet Processing for Cloud Data Center Reliability

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

In cloud computing networks, centralized gateways lead to insufficient reliability and network performance issues due to bandwidth limitations and single-point failures, affecting overall network scalability and reliability.

Innovation Solution

A distributed gateway system is implemented, where each computing node has a local distributed gateway that processes packets, allowing cross-subnet transmission and maintaining network reliability even if a local gateway fails, by using PCIe links and network adapter controllers to modify and forward packets efficiently.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Ease of operation

If a centralized gateway is used to distribute network traffic, then network management is simplified, but network reliability deteriorates due to single-point failure and bandwidth limitations

Engineering Contradiction:
Improvenetwork managementVSAvoidnetwork reliability
Core Design Contradiction:
Ease of operationVSReliability

Solution Approach 1:

The patent divides the centralized gateway function into multiple distributed gateways, with each computing node having its own local gateway. This segmentation eliminates the single-point failure issue while maintaining manageable network operations through localized decision-making at each node.

Inventive Principle:
Principle #1Segmentation

2Device complexity

If a centralized gateway is used for packet processing, then device complexity is reduced, but network performance deteriorates due to bandwidth constraints

Engineering Contradiction:
Improvegateway structureVSAvoidnetwork performance
Core Design Contradiction:
Device complexityVSProductivity

Solution Approach 1:

The patent segments the gateway processing function across multiple computing nodes, allowing each node to independently process packets locally. This eliminates the bandwidth bottleneck of a centralized gateway while distributing the processing complexity across multiple simpler local gateways.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent transitions from a single centralized gateway in one location to multiple distributed gateways across different computing nodes, adding spatial distribution as a new dimension. This dimensional change enables parallel packet processing, significantly improving network throughput and performance.

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

3Ease of operation

If packets are routed through a centralized gateway for cross-VLAN communication, then network control is centralized, but network scalability is limited

Engineering Contradiction:
Improvenetwork controlVSAvoidnetwork scalability
Core Design Contradiction:
Ease of operationVSAdaptability or versatility

Solution Approach 1:

The patent segments the network control function by enabling each computing node to independently perform gateway operations locally. This allows the network to scale horizontally by simply adding new computing nodes with local gateways, without requiring increased capacity at a centralized gateway point.

Inventive Principle:
Principle #1Segmentation

Data Source

PatentUS11831551B2Cloud computing data center system, gateway, server, and packet processing method
Publication Date: 2023.11.28 HUAWEI CLOUD COMPUTING TECHNOLOGIES CO LTD
  • US11831551B2 patent drawing
  • US11831551B2 patent drawing
  • US11831551B2 patent drawing

AI summary

A cloud computing data center system includes a first server, a second server, a cloud management platform, and a switch. The first server includes a first computing node and a first distributed gateway. The first distributed gateway receives a management packet sent by the cloud management platform. The first distributed gateway records network information of the second VLAN. A first virtual machine sends a first service packet that carries service data to a second virtual machine. The first distributed gateway receives the first service packet, generates a second service packet by modifying the first service packet based on the network information of the second VLAN, and sends the second service packet to the switch. A second distributed gateway receives the second service packet forwarded by the switch, and sends the second service packet to the second virtual machine. In this way, network reliability may be improved.