LAN Switch Rapid Fault Recovery via MAC Table Clearing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional Ethernet switches take too long to recover from link failures due to their reliance on aging MAC address table entries and topology change notifications, which can delay network recovery by up to 15 seconds, making it unacceptable for some networks.
Innovation Solution
Implementing a link-loss-learn protocol that clears the MAC address table and momentarily drops links on other ports to propagate the failure, allowing for immediate discovery and relearning of addresses, reducing recovery time to milliseconds.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If the switch uses conventional MAC address table aging and topology change notifications for fault recovery, then network stability is maintained, but fault recovery time is excessive (up to 15 seconds)
Solution Approach 1:
The patent applies preliminary action by pre-configuring the switch to immediately clear the MAC address table upon detecting a link failure, rather than waiting for the conventional aging process. This proactive clearing of the table forces immediate relearning of MAC addresses through the alternative path, reducing recovery time from 15 seconds to approximately 3 seconds while maintaining network stability through the structured discovery process.
2Productivity
If the switch clears the MAC address table immediately upon link failure detection, then discovery process starts immediately, but network traffic disruption increases due to complete table invalidation
Solution Approach 1:
The patent applies segmentation by dividing the network into segments separated by the failed link. When a link failure is detected, the switch clears the MAC address table and sends out discovery packets that are confined to the affected network segment. This segmented approach allows the discovery process to proceed immediately in the affected segment while leaving other segments operational, thereby reducing overall network traffic disruption compared to a complete table invalidation approach.
Solution Approach 2:
The patent uses discovery packets as intermediaries to facilitate the MAC address relearning process. Instead of immediately updating the MAC address table with new information, the switch sends out discovery packets that traverse the network and elicit responses from end stations. These intermediary packets enable the switch to gradually rebuild the MAC address table with accurate information while maintaining network operation, thus reducing traffic disruption.
3Reliability
If the switch propagates the link failure by dropping links on other ports, then fault recovery is accelerated across the network, but link stability is temporarily compromised
Solution Approach 1:
The patent applies preliminary anti-action by proactively dropping links on other ports when a link failure is detected, before the conventional Spanning Tree Protocol would react. This preliminary action prevents potential loops from forming during the recovery process and forces the network to immediately recognize and adapt to the topology change. The temporary link instability is intentionally introduced to accelerate overall fault recovery and prevent more severe network issues.
Data Source
AI summary
One embodiment disclosed relates to a method of fault recovery by a switch in a local area network. A link failure is detected at a port of the switch. In response to the link failure detection, a medium access control (MAC) address table of the switch is cleared. Clearing the address table causes a discovery process to fill the table to begin immediately. In addition, a link on another port of the switch may be dropped to propagate the link failure.