Friday, November 26, 2010

“Adding bandwidth is cheaper than scarcity allocation”?

Here are the numbers courtesy of Benoit Felten and Hermann Wagter - it would be nice to think a regulator somewhere is talking to these guys and working out the same thing?
UPDATE: some commentators suggest that different numbers would be more reliable - the point stands: regulators should be analyzing the issue rather than accepting ISP claims. As a Super JANET user, I would suggest they speak to their architects.

1 comment:

Richard Bennett said...

Felten and Wagter seem to specialize in ludicrous arithmetic.

Not long ago, they offered the claim that Jacobson's Algorithm balanced bandwidth consumption across users, a claim of such absurdity it single-handedly accounted for a 20% increase in hospital emergency room admissions as computer scientists violently rolling about on the floors of their offices upon reading it sustained multiple serious rib injuries.*

The problems with this recent exercise in Onion math are manifold:

1. Since when is the transit network's outer edge (POP to IXP) the most congested part of the Internet?

2. Who says there is always dark fiber in the Internet's most congested region begging to be lit up?

3. Adding capacity doesn't alleviate congestion when there remains a backlog of traffic eager to use it, which is always the case during peak-hour uses elastic applications.

4. Many QoS features are built-in to modern IP routers, and it costs nothing to turn them on.

5. Application demand for bandwidth inherently grows faster than the supply.

6. What about mobile?

Chris, really, these people are the worst kind of quacks.

*poetic license