[110662] in North American Network Operators' Group

home help back first fref pref prev next nref lref last post

Re: Anyone notice strange announcements for 174.128.31.0/24

daemon@ATHENA.MIT.EDU (Leo Bicknell)
Tue Jan 13 11:13:28 2009

Date: Tue, 13 Jan 2009 11:12:49 -0500
From: Leo Bicknell <bicknell@ufp.org>
To: NANOG list <nanog@nanog.org>
Mail-Followup-To: NANOG list <nanog@nanog.org>
In-Reply-To: <9FBD1A9D-4148-4284-80CE-C0F3B9159EEF@sackheads.org>
Errors-To: nanog-bounces@nanog.org


--1yeeQ81UyVL57Vl7
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
Content-Transfer-Encoding: quoted-printable

In a message written on Tue, Jan 13, 2009 at 08:55:40AM -0500, John Payne w=
rote:
> I guess the problem is that AS PATH is overloaded and people forget =20
> that the primary purpose is loop-avoidance.  Everything else is =20
> secondary and much like reading Received headers in SMTP mail, you =20
> really should take everything after your direct neighbor's AS with a =20
> grain of salt.

Actually, I'd suggest your not looking at the primary purpose close
enough.

Loop detection kicks in only when there is a loop.  You see your
own ASN coming back to you.

In the case we're discussing THERE IS NO LOOP.  Someone is mis-using
this feature to control the propagation of routes.  Were the victim
to do a show ip bgp neighbor foo receive-routes and see their own
path they would be reasonable to assume that there is a loop, and
someone is reflecting their own route back to them.

This is a human configuring a device to lie about the loop status
in the network.  That is also the problem with this method, it is
exactly the opposite of what the attribute was meant to convey, and
thus someone on the other end who doesn't know what you're doing
is virtually guaranteed to make the wrong assumption.  You're going
to spin up network engineers looking for routing loops, route leaks,
and other issues if you use this method.

I'd also suggest, as Jared pointed out, there are potential libel
/ trade-dress / slander implications here.  Sending out an AS-Path
of "ASfoo ASbar" is the technological equivalent of the English
statement "foo and bar are interconnected with BGP".  Just because
you hide a false statement in an AS-Path doesn't make it any less
of a false statement.

--=20
       Leo Bicknell - bicknell@ufp.org - CCIE 3440
        PGP keys at http://www.ufp.org/~bicknell/

--1yeeQ81UyVL57Vl7
Content-Type: application/pgp-signature
Content-Disposition: inline

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.7 (FreeBSD)

iD8DBQFJbL17Nh6mMG5yMTYRAoq8AJ9RvR4E60lIqyU2ELPJKDFQtkyiIgCcCkCS
PhiddxNi4XEJK5CF5d17e8I=
=F0vq
-----END PGP SIGNATURE-----

--1yeeQ81UyVL57Vl7--


home help back first fref pref prev next nref lref last post