Rolling Stack Upgrade for Layer 2 Switches
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Critical network systems face significant disruption during software upgrades due to the simultaneous impact on all stack members, which can affect dual-attached hosts even if they are connected through different stack members undergoing an upgrade, necessitating a minimally disruptive stateless upgrade procedure.
Innovation Solution
The implementation of a rolling stack upgrade method that logically segments the stack into upgraded and non-upgraded segments, allowing upgrades to occur one member at a time, with data planes blocked between segments and using cross-stack EtherChannels to maintain connectivity, and enabling MAC persistency to avoid network disruptions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If all stack members are upgraded simultaneously, then the upgrade process is completed quickly, but network disruption and service downtime increase significantly
Solution Approach 1:
The stack is divided into multiple independent segments or members that can be upgraded individually. Each stack member operates independently with its own control plane and data plane, allowing one member to be upgraded while others continue serving traffic. This segmentation enables the upgrade process to proceed without taking the entire network offline.
Solution Approach 2:
The system dynamically transitions stack members between operational states during the upgrade process. Members can be moved from an active traffic-handling state to an upgraded state, and then back to active state. The system adapts its configuration in real-time to maintain service continuity while performing upgrades.
2Reliability
If stack members are upgraded one at a time, then network disruption is minimized, but the overall upgrade time and complexity increase
Solution Approach 1:
The system performs preliminary actions by pre-configuring backup path information and establishing redundancy relationships before the actual upgrade occurs. MAC address persistence is configured in advance, and alternative forwarding paths are prepared, so that when a stack member is upgraded, the network can seamlessly switch to backup paths without manual intervention.
Solution Approach 2:
The system continuously monitors the state of stack members and provides feedback to control the upgrade process. The control plane tracks which members have been upgraded, manages the timing of transitions, and adjusts the configuration based on the current state of the network, enabling automated management of the complex upgrade procedure.
3Reliability
If redundancy is leveraged to minimize traffic disruption, then service continuity is maintained, but the system requires additional configuration and state management
Solution Approach 1:
The system performs self-service by automatically managing the redundancy configurations and state transitions during the upgrade process. The control plane autonomously handles MAC address persistence, updates forwarding tables, and manages the transition of stack members between states without requiring manual reconfiguration, thereby maintaining service continuity while minimizing configuration complexity.
Data Source
AI summary
Stackable Layer 2 switches may be upgraded through a stateless upgrade procedure that minimizes software upgrade complexity while leveraging the intrinsic redundancy of the network to minimize traffic disruption. Disclosed methods may be targeted to platforms that lack support for other non-disruptive upgrade technologies, but that can leverage the intrinsic network redundancy to minimize traffic impact during a stack upgrade.