[178852] in North American Network Operators' Group
Re: distinguishing eBGP from show ip BGP
daemon@ATHENA.MIT.EDU (Mark Tinka)
Wed Mar 11 16:36:42 2015
X-Original-To: nanog@nanog.org
Date: Wed, 11 Mar 2015 22:36:35 +0200
From: Mark Tinka <mark.tinka@seacom.mu>
To: Reza Motamedi <motamedi@cs.uoregon.edu>
In-Reply-To: <CACGuEhEnAVAzm2uhaiswTx3GLppoO0ZoiLVy-jYb3A4VuCyYOw@mail.gmail.com>
Cc: nanog@nanog.org
Errors-To: nanog-bounces@nanog.org
On 11/Mar/15 22:27, Reza Motamedi wrote:
> Thanks again, Mark.
>
> So I guess the short answer is that I can't infer anything about the
> location of physical connectivity having this level of information
> from the control plane.
Not reliably as far as I can tell, no. Someone else can chime in here if
they have another perspective.
One could rely on traceroutes, but that is crude and sometimes
unavailable if you're looking to gather any meaningful data (even though
it might be the closest piece of information you'll get without engaging
with the network operator).
> What if the "Next Hop" is inside the neighbor AS. I know it is a
> rather odd and uncommon case, but can it help?
Control-plane-wise, that could be eBGP Multi-Hop.
Forwarding-plane-wise, that is likely a tunnel operation.
Technically, one can build such a topology. I'd hazard that at least
someone is doing this out there somewhere. However, it's not typical by
any means.
Mark.
