VPLS Multihoming Reactions to Network Failures
VPLS multihoming is designed to protect customer sites from a loss of network connectivity in the event of the following types of network failures:
- Link failure between the CE device and the PE router—BGP
on the PE router is notified when the link goes down. BGP sets the
circuit status vector bit in the MP_REACH_NLRI to indicate that the
circuit is down.
If all of the VPLS local attachment circuits are down, then BGP modifies the down bit in the VPLS advertisement Layer2-Extended-Community to indicate that the customer site is down. When the bit is modified, BGP advertises the route to all of the remote PE routers to notify them that the circuit (and site) is down. Each of the remote PE routers run the BGP and VPLS path selection procedures again and reroute the VPLS pseudowires as needed.
- MPLS connectivity failure to the remote PE router—On
the multihomed PE router, BGP discovers that MPLS cannot connect to
the BGP next hop in the service provider’s network. BGP modifies
the circuit status vector bit in the MP_REACH_NLRI to indicate that
the LSP is down. Once the bit is modified, BGP readvertises the route
to all of the remote PE routers to notify them that connectivity from
the local site to the remote site is down.
The remote PE routers each run the BGP and VPLS path selection procedures again. With the LSP to the original multihomed PE router down, the remote PE routers designate the backup multihomed PE router as the VE device for the multihomed customer site. The pseudowires to and from the remote PE routers are then rerouted to the backup multihomed PE router.
- PE router failure—When either the multihomed PE
router or the BGP process running on it fails, the remote PE routers
detect the expiration of the holdtimer, bring down their peering sessions,
and delete the Layer 2 advertisements from that multihomed PE router.
The remote PE routers each run the BGP and VPLS path selection procedures
again and reroute their pseudowires to the backup multihomed PE router.
Alternatively, the remote PE routers could discover that the BGP next hop, represented by the failed multihomed PE router, is unreachable. For this case, the remote PE routers mark the Layer 2 routes advertised by the multihomed PE router as unreachable. The remote PE routers each run the BGP and VPLS path selection procedures again and reroute their pseudowires to the backup multihomed PE router.
The remote PE routers behave in the same manner if you reconfigure the local preference attribute of the primary multihomed PE router (effectively performing an administrative failover to the backup multihomed PE router). On the primary multihomed PE router, BGP advertises a Layer 2 update with the new local preference attribute to all of the remote PE routers. The remote PE routers each run the BGP and VPLS path selection procedures again and reroute their pseudowires to the backup multihomed PE router.
![]() | Note: In the VPLS documentation, the word router in terms such as PE router is used to refer to any device that provides routing functions. |