No. Each mvpn instance needs a unique p-group address, which results in a mt encap/decap for that mvpn's default tree. In a rosen7 with 3 pes, you have one outgoing and two incoming mdts at a minimum. If desired you can then have usage driven data mdts, but that increases tunnel count at the benefit of sparing Pes that have no interested receivers.
IIRC, we support something like 1023 mdts per TP. Not sure that we have the ability yet to scale with multiple tps. HThs -----Original Message----- From: Stefan Fouant [mailto:sfou...@gmail.com] Sent: Friday, October 16, 2009 12:38 PM To: John Frame Cc: Harry Reynolds; Juniper List Subject: Re: [j-nsp] MVPN Interoperability Issue On Fri, Oct 16, 2009 at 3:17 PM, John Frame <jfr...@juniper.net> wrote: Actually it is possible to support BGP-MVPNs over a non-P2MP capable core by using GRE tunnels. Each PE would require a Tunnel PIC to support both the GRE requirement and the vt interface must be used in the routing instance - however if you are using MXs then it is built in already on the DPCs. If Inclusive trees are used, then P2MP branches are built statically, but automatic trees can be built if using a Selective tree. In the GRE approach, do we have the ability to aggregate multiple MVPNs into a single Inter-PE GRE tunnel? Thanks, -- Stefan Fouant _______________________________________________ juniper-nsp mailing list juniper-nsp@puck.nether.net https://puck.nether.net/mailman/listinfo/juniper-nsp