MLAG Switch In-Service Software Update via Stateful Switchover
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Software updates on switches in a multichassis link aggregation (MLAG) domain typically require the switch to be taken offline, leading to packet drops and decreased network performance due to the need for restarting with updated software.
Innovation Solution
A method for in-service software update (ISSU) is implemented, where a second switch detects the first switch's restart and performs a stateful switch-over (SSO), setting non-peer ports to an error-disabled state, selecting a common MLAG protocol version, and reestablishing the peering relationship, allowing for graceful reinsertion and synchronization of the control plane state.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If the switch is taken offline for software update, then the software version is updated, but packet drops occur and network performance decreases
Solution Approach 1:
The system segments the MLAG domain into two independent switches that can operate autonomously. When one switch needs software updates, the other switch continues to handle traffic, dividing the system's operational continuity across multiple independent units that can be updated sequentially without stopping the overall service.
Solution Approach 2:
The peer-link acts as an intermediary between the two MLAG switches, enabling state synchronization and control plane communication. During software updates, this intermediary channel allows the active switch to maintain traffic flow while the other switch is updated, mediating the transition without interrupting network services.
2Reliability
If the switch is restarted with updated software, then the software version is upgraded, but the switch must be taken offline
Solution Approach 1:
The system performs preliminary actions by maintaining a standby switch that is already synchronized and ready to take over immediately when the active switch is updated. This pre-positioned redundancy eliminates the need for the entire system to be offline during software updates, as the standby switch can seamlessly assume control.
Solution Approach 2:
The system changes the operational state parameters of the switches dynamically. One switch transitions to an updated software version while the other maintains the previous version during the transition period. This parameter differentiation allows continuous operation with mixed software versions, avoiding complete system downtime.
3Reliability
If the switch is restarted, then software update is performed, but packets are dropped resulting in decreased performance
Solution Approach 1:
The system creates a copy of the active switch's functionality in the standby switch, maintaining identical control plane states and forwarding information bases. This copied state allows the standby switch to immediately assume the active role without packet loss when the original active switch is updated, as the copy is already prepared and synchronized.
Solution Approach 2:
The system ensures continuous useful action by maintaining traffic flow through the standby switch during the active switch's update process. The peer-link continuously synchronizes states, and traffic is seamlessly redirected to maintain uninterrupted network service, eliminating the harmful effect of packet drops during software updates.
Data Source
AI summary
A method for performing an in-service software update in a MLAG domain. The method includes restarting the first switch using a second version of software, detecting by a second switch that the first switch is restarting and, in response to the detection, SSO. After restarting the first switch, prior to the first switch performing graceful reinsertion into the MLAG domain, setting all non-peer ports on the first switch to an error-disabled state and selecting a common MLAG protocol version supported by the first switch and the second switch and performing graceful reinsertion, which includes reestablishing a peering relationship between the first switch and the second switch using the common MLAG protocol, and after reestablishing the peering relationship, synchronizing a control plane state of the first switch with the control plane state of the second switch and setting all non-peer ports on the first switch to an active state.