cost of doing business
Joe Abley
jabley at isc.org
Mon Apr 18 15:25:16 UTC 2005
On 17 Apr 2005, at 13:54, Andrew Odlyzko wrote:
> We are talking of two different things here, traffic versus access
> bandwidth.
> It will be a while before the average household generates 5 megabit/s
> traffic.
I don't think that's true. I have seen bittorrent clients running on
machines that have good connectivity (>>typical North American
residential; say 100M access to a data centre on the east coast). With
only moderately-popular torrents (think fan movies like fanimatrix or
starship exeter, a week or so after the slashdot effect has died down)
such a client can easily seed at 10-20Mbit/s.
I think for the average household which contains at least one teenager
with a computer, today, the average household is easily capable of
generating 5Mbit/s, sustained for long periods.
[Widespread use of p2p file sharing can blow transit/peering costs per
port out of the water. Until there's a distribution mechanism for the
kind of content most file sharers exchange that ISPs can participate in
(instead of the current mechanism which ISPs have trouble legally
acknowledging the existence of) it seems reasonable to think the risk
of cost explosion will only get worse; people will continue to share
media around the edge, and ISPs will continue to play whack-a-mole with
the protocols concerned to try and keep their costs under control.]
> Even in Korea and Hong Kong, where the average broadband link is in the
> 5-10 Mbps range, average traffic is about 0.1 Mbps.
Average traffic is not as interesting as peak traffic, I think, from
the consumer's perspective.
For example, if I lived within the permitted catchment of a future
comprehensive BBC video archive, a big last-mile pipe would allow
behaviours that a small last-mile pipe would not, such as pulling video
content on-demand. The fact that I might only pull content a couple of
times a week means the average utilisation might be very low. The
benefit of having the big pipe is fairly clear, however, even with such
low average utilisation.
Joe
More information about the NANOG
mailing list