[2435] in Commercialization & Privatization of the Internet
Billing Systems
daemon@ATHENA.MIT.EDU (Jack Haverty)
Thu Feb 27 09:05:10 1992
Date: Thu, 27 Feb 92 06:01:57 PST
From: Jack Haverty <jhaverty@us.oracle.com>
To: fair@apple.com
Cc: jhaverty@us.oracle.com, com-priv@psi.com
In-Reply-To: "Erik E. Fair" (Your Friendly Postmaster)'s message of Wed, 26 Feb 92 23:42:23 -0800 <23339.699176543@apple.com>
Erik,
Yep, good point. I think there's a few situations like the low-volume T1 that
you can find, and the budget-surprise risk is a real issue. In the old days of
time-sharing, back before the dawn of packets, machine time and disk space were
billed by CPU-minute and kilobyte-month, and a small programming error could run
up a large bill. No one seems to care about that anymore (buy more MIPS! buy
a Superserver!) Maybe bandwidth will be in the same category in the next ten
years. But for now, something analogous to disk/cpu quotas (or *some* kind of
control mechanism on cost) would probably help. There are some carriers who
address this problem by taking the risk on themselves - i.e., they guarantee a
cap on your bill - but I doubt they like that solution. It's simply necessary
to make the sale.
You reminded me of another point - if I'm shopping for network services, it's
tough to figure out what exactly I'm buying. Someone offering a full T1
*connection* for only slightly more money than a partial-T1 *connection* may in
practice deliver less end-end capacity. If I have a T1 connection to a service
network in California and a T1 connection in London, for example, that doesn't
say anything about how much throughput I should expect to actually get between
those two sites. At best, it will be T1 - but it could be 100 bits/second!
Getting providers to guarantee the level of service is as hard as controlling
the budget risks.
Jack
