Seamless Supervisor Restart via Port State Preservation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current mechanisms for software restart in network devices, such as switches, disrupt Layer 2 data plane traffic and cause topology reconvergence, leading to potential loops during linecard or supervisor software upgrades, as all ports are brought down to prevent loop formation.
Innovation Solution
The method involves maintaining the forwarding state of restarting ports and coordinating with neighbor switches to prevent loops, allowing certain ports to continue forwarding during restarts while ensuring no topology changes occur, and deferring state changes until after the restart is complete.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If all L2 ports are brought down during supervisor restart to prevent loop formation, then network loop-free topology is maintained, but Layer 2 data plane traffic is disrupted and topology reconvergence occurs
Solution Approach 1:
The patent segments the supervisor functions into control plane (STP) and data plane (forwarding). During restart, the control plane is restarted while the data plane continues operating through saved forwarding information bases (FIBs), allowing traffic to flow without requiring port state changes that would disrupt connectivity.
Solution Approach 2:
The patent performs preliminary actions by saving the forwarding information base (FIB) and port state information before the supervisor restart begins. This pre-saved state allows the data plane to continue operating seamlessly during the control plane restart without requiring port shutdowns or topology reconvergence.
2Reliability
If all L2 ports are brought down during supervisor restart, then loop formation is prevented, but network connectivity and stability are affected
Solution Approach 1:
The patent introduces an intermediary mechanism where the saved FIB acts as a mediator between the restarting control plane and the ongoing data plane operations. This intermediary allows the data plane to continue using previously learned forwarding information while the control plane restarts, maintaining connectivity without requiring port shutdowns that would affect network stability.
3Adaptability or versatility
If supervisor software is restarted to perform upgrades, then software is updated, but BPDUs cannot be exchanged and topology reconvergence is triggered
Solution Approach 1:
The patent performs preliminary actions by saving the complete forwarding information base and STP state information before the supervisor software restart. This pre-saved state allows the switch to resume STP operations immediately after restart without requiring full topology reconvergence, significantly reducing the time loss associated with software upgrades.
4Productivity
If ports remain in fixed state during restart, then forwarding can continue, but coordination with neighbor switches is required to prevent loops
Solution Approach 1:
The patent implements a feedback mechanism where the restarting switch monitors its neighbors' STP state changes and adjusts its own forwarding behavior accordingly. By detecting when neighbors have converged to new topologies, the restarting switch can safely update its forwarding information without creating loops, enabling continuous forwarding while managing the complexity of neighbor coordination through event-driven feedback.
Data Source
AI summary
Disclosed are methods and apparatus for restarting a first network device having a plurality of ports for receiving and transmitting layer 2 data. The first network device belongs to a network of network devices. When a restart of at least a portion of the first network device is imminent whereby STP is no longer functioning for the first network device during the restart and can no longer alter a spanning tree protocol (STP) state of one or more of the ports and such ports that remain in a fixed state during the restart are referred to as restarting ports, a current state (such as forwarding) of each restarting port is maintained during the restart under predefined conditions. During the restart, each of the restarting ports of the restarting network device portion cooperate with its peer port of a second non-restarting network device that is a neighbor of the first network device so as to prevent layer 2 loops in the network.