Re: [Idr] RFC5065 - Section 5.3

Jeffrey Haas <> Tue, 08 July 2008 13:58 UTC

Return-Path: <>
Received: from [] (localhost []) by (Postfix) with ESMTP id BF7803A6810; Tue, 8 Jul 2008 06:58:13 -0700 (PDT)
Received: from localhost (localhost []) by (Postfix) with ESMTP id 0A4BC3A6810 for <>; Tue, 8 Jul 2008 06:58:13 -0700 (PDT)
X-Virus-Scanned: amavisd-new at
X-Spam-Flag: NO
X-Spam-Score: -2.033
X-Spam-Status: No, score=-2.033 tagged_above=-999 required=5 tests=[AWL=0.232, BAYES_00=-2.599, IP_NOT_FRIENDLY=0.334]
Received: from ([]) by localhost ( []) (amavisd-new, port 10024) with ESMTP id grJHZF5eKiZS for <>; Tue, 8 Jul 2008 06:58:12 -0700 (PDT)
Received: from ( []) by (Postfix) with ESMTP id 90A8F3A6803 for <>; Tue, 8 Jul 2008 06:58:11 -0700 (PDT)
Received: by (Postfix, from userid 1001) id CD3F5244181; Tue, 8 Jul 2008 13:58:20 +0000 (UTC)
Date: Tue, 8 Jul 2008 09:58:20 -0400
From: Jeffrey Haas <>
To: David Freedman <>
Message-ID: <20080708135820.GA31991@slice>
References: <>
MIME-Version: 1.0
Content-Disposition: inline
In-Reply-To: <>
User-Agent: Mutt/1.5.15+20070412 (2007-04-11)
Subject: Re: [Idr] RFC5065 - Section 5.3
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: Inter-Domain Routing <>
List-Unsubscribe: <>, <>
List-Archive: <>
List-Post: <>
List-Help: <>
List-Subscribe: <>, <>
Content-Type: text/plain; charset="us-ascii"
Content-Transfer-Encoding: 7bit


Some of this is my opinion and some my recollection of the events at the
time.  Your mileage may vary.

On Tue, Jul 08, 2008 at 02:44:21PM +0100, David Freedman wrote:
> I would rather have said:
> When comparing routes using AS_PATH length, an implementation MAY
> provide the ability to count CONFED_SEQUENCE and CONFED_SET.
> Looking back through the archives, I see Ilya Varlashkin has commented
> on this previously :

And a few other people, including me.

There is nothing stopping a vendor from providing a knob to accomplish
this very feature.  However, as with everything else regarding BGP path
selection, the important thing is that route selection is done
consistently AS-wide.  As long as that is the case, you can do pretty
much anything you like.

The more widely deployed implementations of BGP ignored confederation
segment length as part of its route selection by default.  Thus this
became somewhat a matter of doing what the big vendors did.

The language thus was written to heavily favor interoperability over
flexibility.  This avoids having to discuss the picky issues of
consistent route selection across the AS within a specific extension.

-- Jeff
Idr mailing list