Node Hot-Swapping via BMC Out-Band Hardware Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The existing node hot-adding process in high-end servers is complex and inefficient due to the need for a BIOS self-test and cold reset, which slows down the hot-adding procedure.
Innovation Solution
A method and apparatus that utilize a baseboard management controller (BMC) to obtain and store static hardware information through an out-band channel, allowing the server to perform node hot-adding without a BIOS self-test and cold reset, simplifying the procedure and improving efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If a BIOS self-test and cold reset operation are performed during node hot-adding, then the node device can be properly initialized and recognized by the platform management software, but the hot-adding procedure becomes complex and operation efficiency decreases
Solution Approach 1:
The patent introduces a baseboard management controller (BMC) as an intermediary device between the node device and the platform management software. The BMC performs the self-test and hardware information collection functions that were previously handled by the BIOS, eliminating the need for cold reset operations. This mediator approach allows the node device to be hot-added without disrupting the existing system initialization process, thereby improving operation efficiency while maintaining reliability.
Solution Approach 2:
The patent replaces the traditional BIOS-based mechanical initialization process with an electronic/BMC-based system. Instead of requiring physical cold reset operations to initialize the node device, the BMC electronically collects hardware information through out-band channels and presents it to the platform management software. This substitution eliminates the complex mechanical reset procedure while achieving the same initialization goal, thus improving hot-adding efficiency.
2Loss of information
If a BIOS self-test is performed after node device power-on, then basic hardware information can be obtained, but the hot-adding procedure becomes complex and time-consuming
Solution Approach 1:
The patent implements preliminary action by having the BMC collect and store basic hardware information from the node device before the hot-adding operation is completed. The BMC performs this information gathering in advance through out-band channels, so when the platform management software receives the hot-adding request, the hardware information is already ready. This eliminates the need for time-consuming BIOS self-tests during the hot-adding process, significantly reducing procedure time while ensuring complete hardware information acquisition.
3Measurement precision
If the platform management software waits for BIOS self-test completion before hot-adding, then hardware information accuracy is ensured, but operation efficiency is decreased
Solution Approach 1:
The BMC serves as an intermediary that independently performs hardware information collection with high precision, then delivers this information to the platform management software. This mediator approach allows the software to receive accurate hardware information without waiting for the traditional BIOS self-test sequence, as the BMC has already captured the necessary data through out-band channels. This resolves the contradiction by ensuring information accuracy while enabling faster hot-adding operations.
Solution Approach 2:
The patent replaces the sequential BIOS self-test mechanism with a parallel BMC-based information collection system. The BMC uses out-band channels to simultaneously gather hardware information without blocking the main hot-adding operation flow. This substitution maintains measurement precision for hardware information while dramatically increasing operation speed, as the information collection no longer serially depends on BIOS initialization completion.
Data Source
AI summary
The present invention discloses a method and an apparatus for node hot-swapping, which simplify the node hot-adding procedure, and improve the operation efficiency of the hot-adding procedure. The present invention includes: obtaining, by a server from a baseboard management controller BMC of a node device to be added, static hardware information of the node device to be added, and storing it into a storage device of the server, where the static hardware information is obtained through an out-band channel by the BMC of the node device to be added; receiving, by the server, a node hot-adding command sent by a user, where the hot-adding command carries identifier information of the node device to be added; obtaining the static hardware information of the node device to be added corresponding to the identifier information from the storage device; and adding, according to the static hardware information, the node device to be added.