VPLS hub & spoke
Hi everyone. I’m trying to understand vpls with a hub & spoke topology
to have the no split-horizon configuration added to them. Not sure if
this is even possible with the autodiscovery option vs using manual.So right now I’ve got four sites setup in a full mesh with the following
configuration on each PE router (ME3600X):
l2 vfi TEST1 autodiscovery
vpn id 3000
service instance 3000 ethernet
encapsulation dot1q 3000
rewrite ingress tag pop 1 symmetric
no ip address
xconnect vfi TEST1
So if I want sites 2, 3 and 4 to not be able to talk to each other
except by going via site 1 what configuration change would I need to
do? I thought that adding “split horizon” to the bridge-domain under
the service instance was the way to go but I’m not so sure.
Ideally, I’d like a scenario where I can have one site as the hub and be
able to take advantage of the autodiscovery for instances when a new
spoke is added to the domain. Granted only the hub is benefiting from
this auto discovery but does that mean that the spokes should be
configured as “manual”?
Thanks for any suggestions.
On the hub device you would have a VFI instance.
The spokes would be configured just as if they were running a point to point xconnect (i.e. xconnect statement on the service-instace)On the hub, under the VFI, you need to configure your pseudowires as “neighbors” – this is where you would put the no-split-horizon statements for the spokes:
l2 vfi aaa manual
vpn id 1
neighbor 18.104.22.168 encapsulation mpls no-split-horizon
neighbor 22.214.171.124 encapsulation mpls no-split-horizon
*auto-route-target*” and play with route-targets import/export to have Hub and Spoke topology like you do for L3VPN’s.