[IP] The Tiger Effect vs byte pricing?

[David Farber <dave@farber.net>]

________________________________________
From: Bob Frankston [Bob19-0501@bobf.frankston.com]
Sent: Wednesday, June 18, 2008 10:05 PM
To: David Farber; nnsquad@nnsquad.org
Cc: 'Kelly Looney'; dpreed@reed.com
Subject: The Tiger Effect vs byte pricing?

There was some speculation on another list as to the cause of an apparent slowdown in connectivity this week and difficulties in reaching Firefox download servers. http://asert.arbornetworks.com/2008/06/the-tiger-effect/ claims that Tiger Woods as the cause. I don’t see how the byte caps would help if this is the real face of congestion. It’s also telling that they considered it merely an annoyance. That would be consistent with my contention that heavy users fighting for a big slice impact each other but don’t impact those needing a relatively small portion of the capacity.

Note that there was some speculation that FireFox downloads were a cause of the slowdown though that traffic wouldn’t have been so concentrated.

The congestion may also be very path-dependent – I didn’t see any general slowdown though I did have a problem getting to FireFox servers. David Reed reported that his connection was down to 1Mbps between MIT and LA. But my home connection – Verizon FiOS – was 10Mbps symmetric Boston/LA. I presume that the peering system exacerbated the problems by limiting the capacity along certain paths to avoid incurring peering charges. The peering system created an anti-networking by partitioning the common infrastructure.

I need to emphasize that much of what I’m saying is speculative and I’m very interested in actual measurements.





-------------------------------------------