[9254] in Commercialization & Privatization of the Internet
an Internet buying coop?
daemon@ATHENA.MIT.EDU (Barry Shein)
Sun Dec 26 19:05:47 1993
Date: Sun, 26 Dec 1993 19:05:11 -0500
From: bzs@world.std.com (Barry Shein)
To: sob@tmc.edu
Cc: karl@mcs.com, tenney@netcom.com, com-priv@psi.com
In-Reply-To: Stan Barber's message of Sun, 26 Dec 1993 17:15:48 -0600 <199312262315.RAA02751@tmc.edu>
>From: sob@tmc.edu (Stan Barber)
>1. Providers sometime argue that only doing flat-rate services is cheaper
>since they spend less overhead generating bills of varying rates to their
>customers. Everyone pays the same thing, so generating bills is easy. I
>don't know if this is still true or not. I can say as a service provider
>within the Texas Medical Center, that it is a heck of a lot easier for me
>to generate a flat rate bill than charge by the incident, but maybe I
>am just not efficient enough to capture the incident charges well. Comments
>from the provider folks?
Well, it depends, but for our system although not zero cost it's not a
big deal.
I think the worst of the metered costs stories come from areas where
metering is not readily available. For example, clocks are easy (punch
each time in and the time out, add it up at the end of the month and
multiply by the rate), packet counts are harder depending somewhat on
what you are using as a "meter" (even routers that maintain these
numbers often don't have easy ways to gather them for billing
purposes.)
Add in any sort of qualification of packets (that is, rules by which
some count and some don't) and it's probably quite a project. Of
course, even that is manageable, I could just say to you the rate per
packet takes into account the notion that some percentage of packets
shouldn't be charged for, etc.
However, if someone came along with a package at a reasonable price to
do all this, say one of the router companies, that cost would fall. So
a lot of the cost is still probably in its roll-it-yourself nature,
the fact that at this point each provider pretty much has to absorb
the cost of developing the methodology for themselves.
Merely generating bills once you have the data in a reasonable form
isn't very hard, certainly not much difference between flat-rate and
metered (computers are wonderously dumb and obedient.)
Most of the concerns previously have revolved around how you get that
data, how to validate it or apply rules, etc. Probably non-trivial
until someone decides to sit down and do it in some centralized way
(like I said, a router vendor, who could then amortize the cost by
charging for it as an extra.) And I'm not even sure it's ultimately a
desireable way to do things, there are other choices.
>2. Some customers operate on flat annual budgets and it is easier to pay
>a flat rate for those customers (and budget for that rate) than to deal
>with the risk in the variability.
Yes, this is certainly a valid issue.
However, we keep going round and round the issue switching at
breathtaking rates from one typical customer to another, entirely
different example.
What the current topic revolved around was individuals who really are
just saying that metered or not they find IP to their homes too
expensive, and perhaps harbor the view that unmetered means cheaper
which is yet to be substantiated tho is also tied into a view, perhaps
sort of justified, that any metering scheme will tend to be reasonable
in cost for a low-to-average volume user, but will discourage the
high-volume user, or something like that.
Very specifically, the issue of setting up residential (neighborhood)
LANs with one point of connection was raised. Sort of reselling, sort
of co-op'ing. You can see how a metered rate tends to make that
harder, one neighbor with a sudden MUD addiction and everyone pays
more (note: they're reacting to exactly the same problems the
providers are!)
Not that this particularly argues that it should be otherwise (at the
charging level), it may just be a problem with the scheme.
So when you cite yourself (Baylor College of Medicine) as a customer
who dislikes metered use it's kind of off the scale of what's being
discussed.
Baylor, as a large entity, I'll guess not only dislikes metering, but
will tend to put enough money on the table to strike a bargain since
their resource needs tend to be a bit more than what these folks are
talking about, $20-$50/month or thereabouts. There would be enough
money involved for the provider to sit down and calculate what various
dedicated, one-time cost resources would really cost and try to work
something out. Baylor also isn't very interested, for administrative
and other pragmatic reasons, in parting out their bandwidth to other
sub-customers, another major issue here.
At well under $100/mo/customer one doesn't have that kind of easy
latitude and generally has to work more in the realm of the
stochastics of shared, relatively expensive resources and how one can
provide sufficient resources to keep a large pool of customers happy,
on average. Since resource usage in such a situation, even things as
simple as tying up a modem, can be dominated by a relatively few heavy
users the idea of metering becomes more attractive.
Put another way, without metering in those situations, and with any
desire to keep customers happy, then the flat-rate has to take into
account the *heaviest* users. So unless that distribution of resource
usage is reasonably flatly distributed then the same level of service
will require significantly more money from the light users to help
subsidize the heavy users.
Sheesh, let's make this real simple: Take a modem pool for SLIP usage.
If on average it takes 100 modems to keep 500 "typical" customers
happy, but then 100 of those customers decide to always be logged in
during peak hours (this *ain't* a stretch, that's what happens) while
others just want to grab their e-mail etc, a few minutes use, then for
them to get in at all you might need to go to 200 modems, a doubling
of a non-trivial resource. *Someone* has to pay for that expansion, in
a flat-rate scheme it will mean that those light users will pay
proportionately more just to get in, the flat-rate had better account
for that, or customers will be unhappy, or the provider will go broke.
-Barry Shein
Software Tool & Die | bzs@world.std.com | uunet!world!bzs
Purveyors to the Trade | Voice: 617-739-0202 | Login: 617-739-WRLD
