[165041] in North American Network Operators' Group
Re: How big is the Internet?
daemon@ATHENA.MIT.EDU (Jay Ashworth)
Fri Aug 16 00:26:36 2013
Date: Fri, 16 Aug 2013 00:26:05 -0400 (EDT)
From: Jay Ashworth <jra@baylink.com>
To: NANOG <nanog@nanog.org>
In-Reply-To: <CC43FB93-C97E-4F49-9E6B-A9B7A10FE20F@ianai.net>
Errors-To: nanog-bounces+nanog.discuss=bloom-picayune.mit.edu@nanog.org
----- Original Message -----
> From: "Patrick W. Gilmore" <patrick@ianai.net>
> > I suspect that, to a first approximation, "traffic which passes through the
> > edge of at least one AS" is probably what most people think of as
> > 'Internet' traffic.
>
> As per my original post to this thread, that would remove all traffic
> from Akamai on-net nodes, Google's GGC nodes, Netflix's on-net Open
> Connect nodes, and many others.
>
> If you are a broadband network in many countries, that is well over
> half the traffic going down your customer's pipes.
>
> I think most people would alter their definition to count that
> traffic.
Ok, "to a zeroth approximation".
That said: it depends on what you're trying to measure, as has been
pointed out before: the entire *point* of edge caching is "to get all
that duplicated traffic 'off of the Internet'," no?
> > As for your DNS question: the interior query isn't, per-se, but the
> > repeated one from your resolver/proxy *is*.
>
> I don't think the type of packet (DNS, HTTP, SMTP, etc. or even TCP,
> IP, ICMP) should matter.
The rest of those are generally not application-level proxied the way
DNS is with most consumer edge NAT routers.
Cheers,
-- jra
--
Jay R. Ashworth Baylink jra@baylink.com
Designer The Things I Think RFC 2100
Ashworth & Associates http://baylink.pitas.com 2000 Land Rover DII
St Petersburg FL USA #natog +1 727 647 1274
