[183] in winnt
Re: Local DHCP Server?
daemon@ATHENA.MIT.EDU (Tom Coppeto)
Thu Jul 2 09:29:00 1998
To: bkaynor@MIT.EDU
Cc: ntpartners@MIT.EDU, jmhunt@MIT.EDU, itpartners@MIT.EDU
Cc: tom@MIT.EDU
In-Reply-To: Your message of "Wed, 01 Jul 1998 14:05:06 EDT."
<3.0.2.32.19980701140506.009d2ca0@hesiod>
Date: Thu, 02 Jul 1998 09:28:27 EDT
From: Tom Coppeto <tom@MIT.EDU>
Bob, I don't have a better answer for you but let me clarify some points.
MIT's cost recovery model for the network breaks down into 2 components.
The costs of equipment, maintenance, renewal and staff to provide a desktop
connection (aka utp charges) and the costs of equipment, maintenance,
renewal, ISP fees, support and staff for the network infrastructure and
core services. Ignoring the wrench of bulk billing which has been under a
fair bit of scrutiny lately, the latter category is recovered on a host by
host basis.
The distinction between the two categories is largely historical but serves
to decouple the number of connections from the number of hosts registered
to that space, a feature MIT is using a lot more lately. We measure the
number of hosts in terms of IP addresses assigned because, again
historically, they have served as the identifier. Dynamic configuration
and future versions of Internet protocols are changing this assumption but
regardless of how the hosts are identified, the model, as it exists today,
is a host by host recovery scheme. DHCP and other address sharing
mechanisms are not relevant in the model.
The goal of the cost recovery system is to recover the cost of providing
the entire network service and to distribute this cost across the user
base. Given this, analyses are underway to re-examine the model and rates
in light of the growing expense of the equipment, the growing expense of
support and the distribution of the hosts (or whether or not using the host
as the metric is a good idea). I also anticipate that the mechanics of
this system will be updated to reduce the administrative overhead and
provide better access to the data. We want to provide networking and not
count the beans all day.
You asked about a usage based billing system, like the one used for tether.
Tether is really a different banana altogether and the usage measured on
dialup lines is not analogous to that measured on an ethernet. One of the
relevant distinctions between the tether system and the campus network is
that one is much bigger than the other. Keep in mind any accounting system
along with FTE would need to be cost recovered out of the same base (take a
look at what the parking accounting system cost and we have a lot fewer
cars).
I hope this helps clarify some things and I'm sorry I don't have a better
answer for you at this time.
- Tom
