[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [NRO comment on ULA 3] Needed changes

Thomas Narten <narten@us.ibm.com> wrote:

|> |and the principle that the charges of 
|> |registration under any given registration model should reflect the costs 
|> |associated with the operation of the service,
|> One of the big benefits of the permanently allocated addresses being
|> discussed is that the costs of maintaining their database can be
|> made extremely small.  The registration charge (if any) should
|> reflect that low cost of operation.
|One thing that needs to be considered (by the entity providing this
|service) is how it can recover costs on an on-going basis over the
|long term (say, 15 years from now). If there are a lot of
|registrations up front (first 5 years) and then they slowly drop off
|to a trickle, who is going to pay for the recurring costs in the years
|that follow?

This question (which has been raised before) seems to be based on the
assumption that current revenue has to be spent or distributed rather
than invested to pay for future costs.

|They may be relatively small, but they are not 0.

Perhaps we should try to quantify them a little better.  My feeling is that
they can be so low that the interest on the money garnered from a reasonable
number of initial registrations will handle the costs indefinitely.  Further
revenue would be gravy.  (I know this begs the question of what is reasonable.)

|What I
|think the RIRs are saying is that they can't provide this service by
|having its costs subsidized by some other RIR income not related to
|these addresses (its membership would rightly object).

The difficulty of distinguishing between this position and the similar
position that all allocations must be as profitable (in a "non-profit"
sense) as current allocations is, IMHO, a reason for concern.

|Indeed, the
|number of organizations willing to take on such a burden is probably
|somewhat limited.

Of course, if the job is offered only to the RIRs we will never know this
for sure...

|So, while the registration fee should be as low as reasonable, it also
|needs to be realistic, in the sense that the long-term funding model
|for the service needs to be stable.
|Looked at in this light, a recurring cost model may not be so

The problem is that the administrative overhead for recurring billing (and
the associated record keeping) is going to swamp the actual cost of maintaining
a record no matter how much you streamline the former.  The RIRs already have
a pretty good idea of this overhead based on their current models.  Isn't
it inevitable that if they handle these new allocations in a similar way they
will arrive at a similar cost/price?  To do otherwise would be to admit that
the existing charges aren't really based on cost recovery.

				Dan Lanciani

IETF IPv6 working group mailing list
Administrative Requests: https://www1.ietf.org/mailman/listinfo/ipv6