Dissimilar Switch Stacking via Capability-Based Master Election
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional stacking of switch information handling systems (IHSs) is limited by the requirement that all switches must be from the same platform, due to differences in CPU endianness, NPU ASICs, and processing capabilities, making it difficult to combine dissimilar switches into a single operational unit.
Innovation Solution
A stacking engine is implemented on each switch IHS to exchange capability information, determine processing system affinities, and elect a master switch based on highest affinities, allowing dissimilar switches from different platforms to be stacked and managed as a single unit with a unified IP address.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If dissimilar switch IHSs from different platforms are stacked, then port capacity and system versatility are improved, but system complexity and compatibility issues worsen due to differences in CPU endianness, NPU ASICs, and processing capabilities
Solution Approach 1:
The patent introduces a stack master IHS that acts as an intermediary between dissimilar switch IHSs in the stack. The stack master exchanges capability information with each switch, determines affinities, and manages the stack operations, thereby mediating the complexity of coordinating dissimilar platforms with different CPUs, NPU ASICs, and processing capabilities
Solution Approach 2:
The patent changes the operational parameters of stacked switches by dynamically determining control plane and data plane processing system affinities based on capability information exchange. The stack master adjusts how different platform switches participate in stack operations based on their specific CPU endianness, NPU ASIC capabilities, and processing strengths, allowing dissimilar switches to operate together efficiently
2Adaptability or versatility
If dissimilar switch IHSs with different CPU endianness and NPU ASICs are stacked, then system versatility is improved, but processing compatibility and operational reliability worsen
Solution Approach 1:
The patent applies local quality by assigning different roles and capabilities to different switches in the stack based on their specific platform characteristics. The stack master identifies which switches have big-endian CPUs versus little-endian CPUs, which have stronger or weaker processing capabilities, and configures them accordingly in the stack hierarchy, allowing each switch to operate in its optimal mode while maintaining overall system reliability
Solution Approach 2:
The patent performs preliminary capability information exchange and affinity determination during stack formation, before actual operations begin. The stack master collects capability information from each switch, determines control plane and data plane affinities in advance, and establishes the operational hierarchy beforehand, preventing compatibility issues from arising during runtime operations
Data Source
AI summary
A switch IHS stacking system includes a plurality of switch IHSs. A least one first switch IHS includes a first processing system and at least one second switch IHS includes a second processing system that is different from the first processing system. A stacking engine is located on each of the plurality of switch IHSs. Following the coupling of the plurality of switch IHSs into a stack and in response to the startup of the plurality of switch IHSs, the each of the stacking engines may exchange capability information with each of the plurality of switch IHSs and determine a control plane processing system affinity and a data plane processing system affinity for each of the plurality of switch IHSs. The stacking engines may then determine a master switch IHS for the stack that has the highest control plane processing system affinity and data plane processing system affinity.