[94913] in North American Network Operators' Group
Re: death of the net predicted by deloitte -- film at 11
daemon@ATHENA.MIT.EDU (Todd Vierling)
Tue Feb 13 01:33:55 2007
Date: Tue, 13 Feb 2007 01:30:04 -0500
From: "Todd Vierling" <tv@pobox.com>
To: "Hank Nussbacher" <hank@efes.iucc.ac.il>
Cc: "Peter Beckman" <beckman@purplecow.com>, nanog@merit.edu
In-Reply-To: <Pine.LNX.4.64.0702130732400.16005@efes.iucc.ac.il>
Errors-To: owner-nanog@merit.edu
On 2/13/07, Hank Nussbacher <hank@efes.iucc.ac.il> wrote:
> I've seen this in action as far back as 1998 and just don't quite grok why
> it never took off.
Let me paraphrase a couple folks who summed it all up very nicely:
"So assuming router state based multicast, how do you bill on that if
the stream is exploded on the opposite end of, or in the middle of, a
transit network?"
The simplified answer of "only as the stream actually transiting the
network" won't fly with most bean counters, because in their eyes,
every packet going through the network should be billed as bandwidth
consumed. Multicast turns that notion inside out, because while
multicast saves bandwidth generally, the bandwidth multiplies as it
transits a for-pay network, meaning that more resources are consumed
and thus ... could be billed for money.
Traditional v4 multicast, then, is unlikely to see deployment outside
of an organiation's own garden network, and you have near zero uptake.
Follow the money, as always. :)
--
-- Todd Vierling <tv@duh.org> <tv@pobox.com> <todd@vierling.name>
