ietf
[Top] [All Lists]

RE: draft-ietf-nat-protocol-complications-02.txt

2000-04-23 00:30:02
they can't get the numbers they need or want, in large measure, due
to the greed of ISPs.

Rather than demonizing ISPs, it's more worthwhile to take
some time to stand in their shoes. Back in the mid '90s,
we faced these same issues in provisioning of small office/home
offices. It was generally much easier (and less expensive from
an administrative point of view) to roll out dialup infrastructure
on a mass scale allocating a single (dynamic address) than it
was to allocate and keep track of address space, with attendant
routing issues.

Early on, the hope was that RADIUS functionality like
Framed-Route and Framed-Routing could enable inexpensive
mass-provisioning of SOHOs connected via inexpensive
RIP-enabled analog and ISDN routers. This functionality
did make it easier to set up statically routed SOHOs.
However, it didn't scale very well as POPs grew to
encompass many access devices, and customers changed
their points of attachment, thus making it increasingly
difficult to administer from a routing point of view.

Support data also told us that those statically routed
customers were many times more likely to require
extensive support. All of this combined to make the
real costs of providing the "routed" service many
times that of giving customers a NAT and a dynamically
assigned address.

Since then, we've made remarkably little progress on
the auto-provisioning problem.

And yes I have a credit card ready, willing and able to purchase service
for an ISP willing to give be a nice block of IP V6 numbers ( 50 will do )
at a reasonable price and the support necessary to implement them.
      ^^^^^^^^^^               ^^^^^^^

The words "reasonable" and "support" do not fit together well, since a
service that requires lots of support isn't likely to be reasonably
priced. What we probably should be shooting for is a service that can
be sold at a reasonable price, but that requires no support at all.