[51351] in North American Network Operators' Group
Re: Measuring BGP routes
daemon@ATHENA.MIT.EDU (Bruce Pinsky)
Fri Aug 23 18:35:56 2002
Date: Fri, 23 Aug 2002 15:31:36 -0700
From: Bruce Pinsky <bep@whack.org>
Reply-To: bep@whack.org
To: Bradley Dunn <bradley@dunn.org>
Cc: Jeff S Wheeler <jsw@five-elements.com>,
"nanog@merit.edu" <nanog@merit.edu>
Errors-To: owner-nanog-outgoing@merit.edu
Bradley Dunn wrote:
> On 23 Aug 2002, Jeff S Wheeler wrote:
>
>
>>On Fri, 2002-08-23 at 15:05, Frank Scalzo wrote:
>>*edited for length*
>>
>>>Most of the tools I have seen have basically done a show ip bgp
>>>collected the whole table and parsed it. I know I wrote a tool like that
>>
>>Is there any reason this is preferable to establishing an eBGP multihop
>>session with the box and receiving the information by that means?
>
>
> Yes.
>
> BGP only passes along the best route. 'show ip bgp' gives you all routes.
> Unless, of course, your vendor supports something like Juniper's
> "advertise-inactive".
>
Actually, BGP will give all routes in the peering session (assuming no
filtering, yada, yada). It won't give you all paths though (subject to your
cavaet above). So depending on which you are looking for, the peering session
idea might be fine.
==========
bep