Network Switch Control Plane In-Service Failover via BMC Replication
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Network switches face challenges in providing high availability and resilience, especially during planned or unplanned events like software upgrades or host processor failures, due to limitations in existing fault tolerance mechanisms that are costly and complex.
Innovation Solution
Implementing an in-service switch-over (ISSO) mechanism where a network switch's functionality is replicated and can be seamlessly transferred from a host processor to a Baseboard Management Controller (BMC), allowing the BMC to take over critical control plane functions, reducing downtime and complexity.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional fault tolerance mechanisms are used to ensure high availability during software upgrades or processor failures, then system reliability is improved, but device complexity and cost increase
Solution Approach 1:
The system segments fault tolerance functionality by separating the control plane into a primary host processor and a secondary BMC processor. Each processor can independently manage network switch functionality, allowing failover without requiring complex integrated redundancy mechanisms throughout the entire system.
Solution Approach 2:
The BMC processor is pre-configured with replicated state information from the host processor before failures occur. This preliminary replication of control plane state enables immediate failover without requiring complex real-time synchronization mechanisms during failure events.
2Loss of time
If state information is replicated from host processor to BMC for failover capability, then switch-over time is reduced, but memory usage and system resources increase
Solution Approach 1:
The system creates a copy of the control plane state information from the host processor memory to the BMC processor memory. This copying approach enables rapid failover by having pre-prepared state information available at the backup processor, avoiding the need for complex real-time state transfer mechanisms during failover.
Data Source
AI summary
An apparatus comprises a first processing device configured to manage, in a first mode of operation, one or more first tasks for a network switch, and to replicate, in the first mode of operation, state information for a network operating system of the network switch, the network operating system running on a second processing device in the first mode of operation. The first processing device is also configured to initiate, in response to one or more trigger conditions, an in-service switch-over from the first mode of operation to a second mode of operation. The first processing device is further configured to suspend, in the second mode of operation, at least one of the one or more first tasks, and to run, in the second mode of operation, at least a portion of the network operating system utilizing the replicated state information.