Hi Fred. You noted:
"My article suggests that the current marketing model, based on peak rate, is wrong. That's the one that gives rise to an oversubscription ratio. Instead, the proper product should be bucket-of-bits-per-month. And that is slightly more amenable to traffic engineering."
This calls to mind a comment made last year by Amory Lovins (and I'm sure others, before and since), concerning the irrelevance of one's momentarily lowering their electric usage on the total cost of power generation (and its environmental effects), since supply will be generated at a relatively fixed rate and cost, anyway. [Citing from somewhere near the beginning of his talk here: sic.conversationsnetwork.org ]
Of course, that last claim doesn't properly take into account the ability to dynamically allocate supplies from sources other than those tied down to large generation plants, and further disregards the costs associated with appropriately sizing electric transmission and distribution facilities (at the substation and feeder levels) downstream.
If the latter (downstream facilities) aren't sized properly, then, in the "broadband" scenario, at least, which includes the last mile provider's switch/cmts/hub and local loop ratings, then the size of the larger bit bucket residing upstream doesn't matter all that much anyway -- except for periods of very low aggregate use, when an individual user might have access to the entire supply.
So where, in your example, does the larger bucket of bits reside? Is it at the last mile provider's CO/HE/Hub (substation), or farther upstream in the core? A rhetorical question, of course, but it casts a slightly different light on some of the factors involved, from which one might be inspired to think of alternative topologies.
FAC
------ |
