A cost-based model

[Howard C. Berkowitz]

Trembling as I make my first post to this list, may I make a suggestion?
As I see it, much of the financial concern with registry fees are regarding
them as a price paid by address assignees.  There is concern that these
prices may be too expensive for small companies, and/or will be passed to
end users and raise end user prices.

I will make the assumption that there is a value to having registries and
control of addressing; I recognize some people here believe that this is
artificial and part of the CIDR Conspiracy.  Even if there were such a
conspiracy, I don't know how any routing system could work if there were no
bodies to assure unique address assignment.

So I suggest, in a completely serious, non-sarcastic way, what I think
might be a useful reality check.  RFC2050 certainly does describe the
functions of a registry; there's real world experience to extend it.

I propose that some of the people concerned that current price proposals
come up with a cost-based proposal, and see how it reconciles with the
price proposals.  In other words, try to cost out some of the things a
registry will do.

As a starting point, look at cost factors in address allocation.

   There must have been prior work to set up a data base for the aggregated
address space, to set up allocation logic and operational procedures, etc.
These need to be amortized over the number of assignments the registry will
make, along with the training costs of assignment technicians.

   Assume assignment requests arrive electronically.  There will be a cost
of connectivity such that they are received. There will be a cost of
maintaining public servers with assignments.

   Address assignments will need to be reviewed by a human being with
certain knowledge.  How long will it take to review the average request
with no complications or errors?  How frequently will there be errors?  For
the number of allocation requests reasonably expected, how many first-level
and higher-level people are needed to process them?

   What skill level do these people have and what is the market price for
those skills?  What administrative, HR, managerial, etc. support do they
need?  In today's world, the registry must assume it eventually will be
sued, and budget for legal cost both preventive and reactive.

Going through such an exercise will give a first approximation on the cost
of managing allocations, ignoring legitimate planning, policy and
coordination time.  But the cost of allocation, with reasonable
administrative and policy overhead, should have some relationship to the
price for address allocations.


Howard