Hi guys,
my customer (it's an ISP) is requesting a L2VPN solution that satisfy these requirements:
1) His clients, single homed via ME3600x acting as MPLS PE, want to be able to realize L2VPN p2p and p2mp among their remote sites spread geographically around the ISP MPLS network
2) ISP's clients' services, hosted in the ISP Data Center, must have the opportunity to be reached as L2VPN from remote sites (L2 extension)
3) The data Center has a couple of 7609 facing the MPLS backbone in dual homing (as shown in the attachment)
What would you suggest:
1) A mix solution of PWs via xconnect among PEs for p2p L2VPN and VPLS with Bridge Domain introduction for p2mp L2VPN?
2) A ONE solution with VPLS that use BGP for autodiscovery and BGP/LDP for signaling for both p2p and p2mp PWs?
3) For the L2VPN connection towards the couple of 7609, which redundancy approach do you suggest? MG-LAG or simply the primary and backup p2p PWs leaving from remote PE sites towards the couple of 7609?
"p2p xconnect" would avoid to manage for PEs the MAC add learning but i would lose the opportunity to make easier and more scalable the configuration via BGP autodiscovery and BGP/LDP signaling avoiding to have to manage the single p2p PWs to be configured each time a new client requesting that service is appearing into the scene.
Vice versa, "VPLS" with bridge domain and so on, would require to PEs to memorize MAC add... but easier to manage as configuration.
What is your opinion?
Thanks,
Mario
