[ppml] IPv4 Soft Landing - Discussion andSupport/Non-SupportRequested

[michael.dillon at bt.com]

> All I'm going to say is that if you ignore the enforcement 
> aspect you risk creating a policy that is impossible to implement
> - and as a result, will not be implemented.  That is what I 
> am saying your doing here.

And you can't enforce a policy if you don't know what is going on. This
is the reason why I think that the focus should be on things like
surveys, reporting, auditing of past allocations. This can make things
better all by themselves. For instance a survey that requires the
signature of the CFO will raise the profile of IP addressing in all
organizations using IP addresses. Reporting could be designed to give a
clearer idea of whether or not an organization actually has good
management controls for IP addresses internally. And auditing will help
organizations find unused addresses in their organization and will also
help them improve their controls.

In previous companies I have worked at, we stopped giving out new IP
addresses and miraculously, we were still able to install new customers.
The ops people found a few assignments that were bigger than they needed
to be and clawed them back. And there were some disconnections whose
addresses had not been returned to the pool. In another company, I
remember a program to get rid of the $2 million that was being paid
EVERY MONTH for circuits (mainly access circuits) that we no longer
used. It seems that cancelling circuits was a really low priority.

A significant number of IP addresses are allocated to companies who are
large enough to suffer from these types of internal process problems. By
helping companies to address these problems in a non-adversarial way, we
can accelerate deployment of IPv6, at least to the test-lab and trial
stages, and we can find enough IP addresses to give another 6 months to
a year after ARIN runs out.

> Why don't you go ask them right now to look at your policy 
> and tell you if it was implemented, would they be following 
> up a year later with the requestors to make sure they had 
> actually carried out the 50% demonstrated requirement in 
> Phase 0 & 1, the demonstrated 75% requirement in Phase 2 and 
> the demonstrated 90% requirement in Phase 3.  

You highlight another major problem with this proposal. Implementation
of such a complex policy will not be done quickly. If it takes 18
months, then we are already in a different part of the IPv4 exhaustion
endgame and this kind of soft-landing simply won't work.

--Michael Dillon