> Gert Doering > Sent: Friday, October 18, 2019 10:45 AM > > Hi, > > On Fri, Oct 18, 2019 at 11:40:53AM +0200, Mark Tinka wrote: > > On 18/Oct/19 09:15, Gert Doering wrote: > > > > > I could see very special cases where it would be necessary, but that > > > would need to be a non-default-enabled switch. > > L2PT would be a use-case, but as you state, not typically standard. > > If I understand "L2PT" right here, this is the classic "EoMPLS" (in Cisco > language) or "CCC" thing, as in "transparently connecting exactly > *two* ethernet ports together, over MPLS or L2TPv3 or ... transport"? > > If yes, is this something people do over EVPN? > > I'm asking because I'm trying to understand options - we see EVPN as a tool > for "I need a learning bridge in the middle, with more than two endpoints", > while EoMPLS/CCC is "I want to connect exactly two endpoints, and the > middleboxes must be fully transparent, including STP forwarding". > Even more importantly if doing virtual L2 pipes one doesn't want to be involved in any old crap that's flowing through those pipes -including mac learning, So yes I guess EVPN can be used also for p2p pipes (to say unify your technology suite across the carrier ethernet services) but in that case the bridges at both ends need to have mac learning disabled and since I haven't played with this kind of setup I'm not sure what info I'd need to exchange over the mp-bgp sessions or how to put static bridge forwarding rules in order to make packets flow bidirectionally. THEN I'd be concerned with ok and now how do I forward BPDUs over this mac-less setup.
So yes BPDUs forwarded to customer ports on a local bridge -may be ok to allow customer's spanning tree work Bot no way how in mp2mp setup one would ever need BPDUs forwarded to other PEs. (if tis required it's a bad design). adam _______________________________________________ juniper-nsp mailing list juniper-nsp@puck.nether.net https://puck.nether.net/mailman/listinfo/juniper-nsp