Hi all.
I have some problems to understrand vpls over mpls and I have some doubts.
Are these instances true?
1. mpls need ldp to exchange lables
2. vpls need mpls to be esthablished
3. the same vpls tunnel could be esthablished between more than one router (it is different from eoip)
4. the label added from vpls could be distributed by LDP or BGP (using l2vpn)
5. using bgp, if I want to configure the same vpls tunnel between a lot of routers, I need only to make ONE configuration each router (the full mesh is automatically established)
6. the only advantage to use BGP is the possibility to ethablish a full mesh automatically
7. If in an ospf network I want to use also bgp signalled vpls, I can connect each router with the adiacents by a bgp link

Thanks

Its like YES to all. But I would assume a VPLS to each router from a concentrator etc, and have the VPLS into a bridge interface. (maybe with horizon)

Its quite simple.

make a loopback, and have a loopback ip.
Have a link net to next router, and put the interfaces to this l2 net. Also enable MPLS to this interfaces. Have ospf up running (distributing the loopback ip)
then make a BGP connection using vpn4 and l2tp-cisco. to loopback ip. Add vrf and or confederation at your need.
If all is up running, just start to "publish" VPLS tunnels by connecting to each routers loopback IP's.

and next, and next, etc. The all should be in auto.