re Who Confirms The Accuracy (or is it precision) Of ISP Usage [sic] Meters?

[David Farber <dave () farber net>]

Begin forwarded message:

From: Stan Hanks <stan () colventures com>
Date: January 13, 2010 4:47:48 PM EST
To: dave () farber net
Subject: RE: [IP] re Who Confirms The Accuracy (or is it precision) Of ISP Usage [sic] Meters?

Bob Frankston says:
> The deeper issue is our willingness to accept the bad metaphor that leads us to think we are using up the "Internet" 
> as if we were consuming electricity.
Bob’s wrong – if you’re using bandwidth on a path, I can’t use it on the same path. While we’re not talking about a 
finite natural resource, like oil, or similar, we’re talking about what I’ll characterize as financially limited finite 
resource: bandwidth.
Let me be clear: statistical multiplexing, packet switching and queuing mean that if I attempt to push bits down a 
path, sooner or later, they’ll get there. The reality is that queues are memory intensive (expensive) and that routers 
are space and power intensive and don’t scale well (really expensive) and that users measure “bandwidth” not in 
absolute terms, but in terms of how it feels – it’s either fast (high bit rate, low latency) or slow (lower effective 
bit rate, high latency) or unusable (really low bit rate or really high latency or both). Hell, I once sent IP packets 
over pulses in drilling mud at 3bps (yes, THREE bits per second) so you CAN move data at really low speeds – but for 
most applications, being seen as “useful” takes a lot more bandwidth.
Sure, you can make more, but that takes MONEY. Ummm. Got any laying around? I sure don’t, at least in the quantities it 
takes to build national scale bandwidth…. Oh, I can probably raise some, but only if I can prove that there’s a 
reasonable return on doing so – which requires that I somehow or another get more money from the users that I already 
have, or find users that I don’t have. If I just give the “new” bandwidth to the old customers at the old price, or 
WORSE at a lower price, then sooner or later, I’m going to have investor issues…
We CAN NOT get around this core issue. As much as we’d like it to go away, you only get X amount bandwidth for Y amount 
of spend. Sure, X goes up if you delay how long you wait to spend Y, but at the end of the day, no one has ever sold, 
and is unlikely to ever sell, gear that I buy today that will let me scale up my bandwidth perpetually with no addition 
expense other than “routine maintenance”.
So, to me, the dialog is this: how much degraded performance are consumers willing to tolerate before they’re willing 
to spend more money, and what can be done to move both that tolerance level and the amount of money they’re willing to 
spend to a point where you get equilibrium?
The counterpoint: over-the-air broadcast TV. It runs on the basis of advertising. Google is similar as are others. If 
users aren’t willing to pay for what needs to be paid for, are they at least willing to accept the necessity of that 
sort of model??
Either way, someone has to pay, or you just flat run out of usable bandwidth sooner or later.
 




-------------------------------------------
Archives: https://www.listbox.com/member/archive/247/=now
RSS Feed: https://www.listbox.com/member/archive/rss/247/
Powered by Listbox: http://www.listbox.com