Suppose that a failure occurs at point 1 in the following figure.
The EAPS (Extreme Automatic Protection Switching) master detects the topology change (either through a failure notification from a node on the ring or through a hello timeout), and it unblocks the port on the protected VLAN (Virtual LAN) at point 2. The Dist 2 node now connects to the VPLS through Core 2 instead of through Core 1.
When the topology changes either on an access ring or the shared port link, the path used to reach customer devices can change. For example, in the following figure, the path that Dist 2 takes to reach other parts of the VPLS network changes following the failure on the access ring at point 1. Prior to the failure, Dist 2 used Core 1 to reach the VPLS network. Following the failure, Dist 2 accesses the VPLS network using Core 2.
When the EAPS master detects a topology change, it sends a flush FDB (forwarding database) message to its transit nodes. The transit nodes re-learn all the MAC addresses on the ring. However, this flush FDB message is not propagated over the VPLS network. As a result, Core 3 in the following figure still expects to find Dist 2 through the PW between Core 3 and Core 1. Any traffic destined for Dist 2 that is sent to Core 1 will not reach its destination.
To correct this problem, EAPS informs VPLS about any received EAPS flush FDB messages on both the controller and the partner nodes, and VPLS performs a local flush of any MAC addresses learned from the originating nodes. In this example, the EAPS processes in both Core 1 and Core 2 notify VPLS because neither node knows where the access ring is broken. The VPLS services in Core 1 and Core 2 send flush messages to the other VPLS nodes.
Print
this page
Email this topic
Feedback
View PDF
Download EPUB