[address-policy-wg] A comment on 2008-03 & 2007-09

> I disagree with you. RIR YES would need a policy for doing 
> what you are suggesting.
> In fact, the policy proposed is a particular case of what you 
> are suggestign.
> 
> When there are only 5 /8s IANA ask all the RIRs what would be 
> your future (short term) needs. And obviously the answer 
> would be "At least one /8".

If I were IANA then that would not be an acceptable answer.

At some point, IANA will be allocating large percentages of
the global free supply of IPv4 addresses to RIRs and they will
have to satisfy themselves that they are not treating the other
5 RIRs unfairly. In the scenario that you describe, you expect 
IANA to give out 20% of the total IPv4 address space just because
you claim that you need it.

This is not sufficient. At a minimum, you should justify your request
by showing the full status of all your previous IPv4 allocations
and this information should be made public. After that, all other
RIRs and the public should have an opportunity to file comments with
IANA. Finally, after considering all the comments, IANA should decide
how many IP addresses to allocate.

IANA would be within its rights to begin doing this tomorrow, not 
waiting for the last 5 /8s. 

My main concern with all these types of proposals is that they
should not cause an IPv4 address shortage to happen sooner than
it would happen without intervention. Some proposals, like the one
which locks up the last 5 /8 blocks would cause shortage effects
sooner, and are therefore bad. 

However, if IANA/ICANN starts taking its responsibility seriously
and  applies more scrutiny, in public, of all RIR allocations, then
we would be unlikely to cause shortage effects earlier. But IANA and
ICANN need to take the initiative on this, not wait for the RIR/NRO
two to three year policy cycle. The criteria for RIR allocations can
and should vary on a case by case basis.

For one thing, I'm aware that the RIRs have a lot of addresses tied up
in "reserved" status which should be taken into account when analyzing
the overall picture. I'm also aware that by taking up the slack, we may
extend IPv4's lifetime by another year or two but when we do run out it
will be like hitting a brick wall. Nevertheless, companies which cannot
implement IPv6, within the next two to three years, are failing in their
duty to shareholders and customers. I don't think we should be trying
to protect these companies in any way, just extend things as far as
we can into the future to give time for the network operator and vendor
project cycles to deliver results.

In the current world of business, project cycles are longer than they
used to be back in the early 1990s. Many people don't like this but 
we have to live with it, since we all depend on other companies
to play their role in IPv6 deployment. We need better support on the
hardware, we need upgraded OSS/NMS software, firewalls, load balancers,
and consumer broadband gateways. In addition we need scalable v6/v4
translation clusters. Any policy changes, and behavior changes, need
to support all these companies in providing the necessary upgrades
within
their plodding project cycles.

I don't think that the proposals currently before any of the RIRs do
this,
and I don't believe that the RIR/NRO policy processes work quickly
enough
to deal with this issue. Only ICANN and IANA can take the needed action
in the required amount of time. Fortunately ICANN is also an open and
responsive organization so we have a chance to fix this in their forums.

-- Michael Dillon