Boot File Network Configuration for HPC Startup

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

In server clusters, the current dynamic addressing method using DHCP protocol requires two separate requests for network configuration, leading to performance penalties and network noise, especially in High Performance Computing (HPC) environments where network availability is critical.

Innovation Solution

The method involves the dynamic addressing module allocating a network address and transmitting it to the boot file supply module, which includes the IP address as a parameter in the boot files, eliminating the need for a second DHCP request during the boot process by providing network configuration information directly to the operating system.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Adaptability or versatility

If dynamic addressing using DHCP protocol is implemented for network configuration, then network configuration flexibility is improved, but network performance deteriorates due to multiple broadcast requests

Engineering Contradiction:
Improvenetwork configuration flexibilityVSAvoidnetwork performance
Core Design Contradiction:
Adaptability or versatilityVSProductivity

Solution Approach 1:

The patent applies preliminary action by including the network configuration information (IP address, subnet mask, gateway) directly in the boot file that is provided to the computing device during the boot process. This allows the operating system to obtain network configuration information in advance without needing to send a second DHCP request, thereby resolving the technical contradiction by maintaining configuration flexibility while eliminating performance degradation from multiple broadcast requests

Inventive Principle:
Principle #10Preliminary action

2Reliability

If multiple DHCP requests are broadcast during startup, then network configuration reliability is improved, but network noise increases and performance decreases

Engineering Contradiction:
Improvenetwork configuration reliabilityVSAvoidnetwork noise
Core Design Contradiction:
ReliabilityVSObject-generated harmful factors

Solution Approach 1:

The patent embeds the network configuration information (IP address, subnet mask, gateway) directly in the boot file that is provided to the computing device during the boot process. This preliminary inclusion of configuration data eliminates the need for a second DHCP broadcast request, thereby maintaining configuration reliability while significantly reducing network noise and performance degradation

Inventive Principle:
Principle #10Preliminary action

3Productivity

If static network configuration is used instead of dynamic addressing, then network performance is improved by eliminating DHCP requests, but device complexity increases due to manual configuration requirements

Engineering Contradiction:
Improvenetwork performanceVSAvoidconfiguration complexity
Core Design Contradiction:
ProductivityVSDevice complexity

Solution Approach 1:

The patent applies preliminary action by pre-configuring the network configuration information (IP address, subnet mask, gateway) directly in the boot file that is automatically provided to the computing device. This approach eliminates the need for manual static configuration or multiple DHCP requests, thereby improving network performance while keeping the deployment process simple and automated

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The boot file supply module automatically provides the network configuration information to the computing device through the boot file, eliminating the need for manual configuration by system administrators. The computing device receives all necessary network parameters automatically during the boot process, reducing both configuration complexity and network noise

Inventive Principle:
Principle #25Self-service

Data Source

PatentEP2449751B1Method of starting up a computing device in a network, server and network of computing devices for the implementation thereof
Publication Date: 2015.04.29 BULL SA
  • EP2449751B1 patent drawingFigure 1
  • EP2449751B1 patent drawingFigure 2
  • EP2449751B1 patent drawingFigure 3

AI summary

This method of starting up a computing device (12) in a network comprising a dynamic addressing module (14) and a module (16) for providing start-up files (32) for said computing device comprises the following steps; allocation (204) to the computing device (12), by the dynamic addressing module (14), of an address in the network, transmission (208), by the computing device (12) to the module (16) for providing start-up files, of at least one request (ReqB) for start-up files (32), and provision (210), by the module (16) for providing start-up files to the computing device (12), of these start-up files in response to said at least one request (ReqB). Moreover, said at least one request (ReqB) comprising the address of the computing device (12), allocated by the dynamic addressing module (14), the start-up files are provided by the file server with this address.