[e2e] Re: economic models for access bandwidth

Xin Wang xwang at ctr.columbia.edu
Thu Apr 19 07:40:09 PDT 2001


Actually I am always interested in knowing the best way to provision the
network when most of the access links are no longer the bottleneck.

On Wed, 18 Apr 2001, Barney Wolff wrote:

> I'm astonished to see replayed on end2end the discussion that
> occupied com-priv 7-8 years ago, with no new arguments or ideas
> from either side.
> 
> Having worked for an ISP, I have to side strongly with Ran here.
> Real world experience says that you cannot stay in business if
> you offer, at higher cost, services that 99% of your subscriber
> population do not use.  Nor can you survive if your whole base
> gets black-holed or banned from IRC because of a few abusers.
> 
> The residential traffic model drives how the network is physically
> provisioned, so it may well be impossible to accomodate a heavy
> user even if it were easy to charge him more.
> 
> TCP copes with the real world.  Don't expect ISPs to behave as
> though they lived in an ideal one.  Even the realists are going
> under or getting taken over.
> 
> Barney Wolff
> 




More information about the end2end-interest mailing list