Distributed BMC Management for Multi-Node Server Complexity

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

In multi-node Information Handling Systems (IHS), managing multiple server instances on a single motherboard increases complexity, and existing solutions require multiple high-cost SOCs to support full IPMI stacks, leading to increased cost and complexity.

Innovation Solution

Implementing a distributed baseboard management system with a reduced IPMI stack on each node's integrated management controller (iMC) and a more complex stack on the baseboard management controller (BMC), allowing for efficient consolidation of management operations across multiple nodes via a low pin-count interface, thereby reducing overall cost and complexity.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If full IPMI stacks are implemented on each node's SOC in multi-node systems, then management capabilities are maintained, but cost and complexity increase

Engineering Contradiction:
Improvemanagement capabilityVSAvoidsystem complexity
Core Design Contradiction:
ReliabilityVSDevice complexity

Solution Approach 1:

The IPMI management functionality is segmented into two parts: a simplified iMC stack on each node's SOC handling basic monitoring and communication, and a full BMC stack on dedicated management processors handling complex management operations. This segmentation allows each node to have minimal management overhead while maintaining full management capabilities through the centralized BMC.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The complex IPMI stack is extracted from individual node SOCs and consolidated into centralized BMCs. Each node's SOC contains only essential management functions (iMC), while the full IPMI processing capabilities are taken out and placed in shared BMC resources, reducing per-node complexity while preserving management functionality.

Inventive Principle:
Principle #2Taking out (Extraction)

2Reliability

If multiple high-cost SOCs are used to support full IPMI stacks on each node, then management features are preserved, but cost increases

Engineering Contradiction:
Improvemanagement feature completenessVSAvoidcost
Core Design Contradiction:
ReliabilityVSQuantity of substance

Solution Approach 1:

The BMC is designed as a universal management processor that serves multiple nodes simultaneously. A single BMC can manage multiple iMCs across different nodes, providing full IPMI functionality for each node without requiring dedicated full-featured SOCs at each node. This multi-functionality reduces the total number of expensive management processors needed.

Inventive Principle:
Principle #6Universality (Multi-functionality)

Solution Approach 2:

Multiple management functions and node control capabilities are merged into a single BMC. Instead of having separate full IPMI stacks on each node, the patent combines management resources at the system level, allowing one BMC to handle IPMI processing for multiple nodes, thereby reducing overall system cost while maintaining feature completeness.

Inventive Principle:
Principle #5Merging (Combining)

3Device complexity

If simplified management architecture is used on each node, then cost and complexity are reduced, but management effectiveness may be compromised

Engineering Contradiction:
Improvenode complexityVSAvoidmanagement effectiveness
Core Design Contradiction:
Device complexityVSReliability

Solution Approach 1:

The iMC on each node acts as an intermediary between the simplified node SOC and the full-featured BMC. The iMC handles local sensor monitoring, event detection, and basic communication, while the BMC provides comprehensive management capabilities. This intermediary approach allows nodes to have simple architectures while maintaining effective management through the BMC's full IPMI stack.

Inventive Principle:
Principle #24Intermediary (Mediator)

Solution Approach 2:

The management architecture transitions from a horizontal distribution model (full IPMI on each node) to a vertical hierarchy model with iMCs at the node level and BMCs at the system level. This dimensional change in architecture allows simplified node designs while preserving management effectiveness through the layered approach, where the BMC provides comprehensive control from a higher system level.

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

Data Source

PatentUS9767067B2Baseboard management systems and methods with distributed intelligence for multi-node platforms
Publication Date: 2017.09.19 DELL PROD LP
  • US9767067B2 patent drawing
  • US9767067B2 patent drawing
  • US9767067B2 patent drawing

AI summary

Baseboard management systems and methods with distributed intelligence for multi-node platforms. In an illustrative, non-limiting embodiment, an Information Handling System (IHS) may include a plurality of modules, each of the plurality of modules including a plurality of nodes, each of the plurality of nodes including a system-on-chip (SoC), each of the plurality of SoCs including an integrated management controller (iMC), each of the plurality of iMCs configured to implement a first intelligent platform management interface (IPMI) stack having a first architecture; and a plurality of baseboard management controllers (BMCs), each of the BMCs disposed on a corresponding one of the plurality of modules, each of the BMCs coupled to the plurality of iMCs on the corresponding one of the plurality of modules, each of the plurality of iMCs configured to implement a second IPMI stack having a second architecture different from the first architecture.