Re: [sidr] WGLC: draft-ietf-sidr-bgpsec-reqs

Eric Osterweil <> Thu, 10 November 2011 18:46 UTC

Return-Path: <>
Received: from localhost (localhost []) by (Postfix) with ESMTP id D648E21F84D4 for <>; Thu, 10 Nov 2011 10:46:44 -0800 (PST)
X-Virus-Scanned: amavisd-new at
X-Spam-Flag: NO
X-Spam-Score: -6.47
X-Spam-Status: No, score=-6.47 tagged_above=-999 required=5 tests=[AWL=-0.098, BAYES_00=-2.599, RCVD_IN_DNSWL_MED=-4, SARE_SUB_OBFU_Q1=0.227]
Received: from ([]) by localhost ( []) (amavisd-new, port 10024) with ESMTP id xcPkwTQ44jdf for <>; Thu, 10 Nov 2011 10:46:44 -0800 (PST)
Received: from ( []) by (Postfix) with ESMTP id D50D121F849C for <>; Thu, 10 Nov 2011 10:46:36 -0800 (PST)
Received: from ([]) (using TLSv1) by ([]) with SMTP ID; Thu, 10 Nov 2011 10:46:38 PST
Received: from ( []) by (8.13.6/8.13.4) with ESMTP id pAAIkRb3010412; Thu, 10 Nov 2011 13:46:27 -0500
Received: from ([]) by with Microsoft SMTPSVC(6.0.3790.4675); Thu, 10 Nov 2011 13:46:26 -0500
Mime-Version: 1.0 (Apple Message framework v1084)
Content-Type: text/plain; charset="us-ascii"
From: Eric Osterweil <>
In-Reply-To: <>
Date: Thu, 10 Nov 2011 13:46:26 -0500
Content-Transfer-Encoding: quoted-printable
Message-Id: <>
References: <> <> <p06240808cad5c4d268eb@> <> <p06240804cad81a9e4485@> <> <> <> <> <> <> <>
To: Christopher Morrow <>
X-Mailer: Apple Mail (2.1084)
X-OriginalArrivalTime: 10 Nov 2011 18:46:26.0923 (UTC) FILETIME=[0D7283B0:01CC9FD9]
Cc: "Sriram, Kotikalapudi" <>, sidr wg list <>
Subject: Re: [sidr] WGLC: draft-ietf-sidr-bgpsec-reqs
X-Mailman-Version: 2.1.12
Precedence: list
List-Id: Secure Interdomain Routing <>
List-Unsubscribe: <>, <>
List-Archive: <>
List-Post: <>
List-Help: <>
List-Subscribe: <>, <>
X-List-Received-Date: Thu, 10 Nov 2011 18:46:45 -0000

On Nov 10, 2011, at 1:41 PM, Christopher Morrow wrote:

> On Wed, Nov 9, 2011 at 3:37 PM, Eric Osterweil <> wrote:
>> Hey Sriram, Russ, and Jakob,
>> Thanks for the #s.  I think I get the general notion that adding n updates per day per prefix equals (n * #prefixes)/1. :)  I guess my question was kinda vague, sorry.  Upon reexamination, I see that I said "overhead" without being specific.  Since we can use the updates that are generated today to measure how much (for example) bandwidth is already needed, can we calculate how much extra bandwidth universal deployment would mean?  Also, perhaps this would be most informative in the form of a ratio (i.e. a factor of $x$ increase).  That way, when people look at events like the one that the "General Internet Instability" thread that just happened on NANOG refer to, they can gauge the update amplification that was seen against what _would_ be seen given bgpsec.  I think this actually kind of came up on nanog, so it seems like maybe it would be a relevant thing to look at here?
> is the 'bandwidth' of the bgp protocol in the wire an actual concern?
> (at some point the discussion point came up ~1yr or more ago, but was
> discarded as not relevant given circuit sizes and bandwidth from link
> -> RP/RE/etc, so I'm genuinely curious about this)

I think it is just a concrete way to relate the amount of data being consumed today, to what may be needed tomorrow.  It isn't so much that 1 byte = good and 10 bytes = bad.  More that in trying to quantitative compare two behaviors, finding a common reference point seems like a good start, imho.  I think a meaningful ratio is more useful, but it just needs something to compare.