[arin-ppml] Requesting Feedback: Draft Policy ARIN-2014-22: Removal of Minimum in Section 4.10

[Owen DeLong]

> On Jan 10, 2015, at 10:51 , William Herrin <bill at herrin.us> wrote:
> 
> On Sat, Jan 10, 2015 at 3:15 AM, Owen DeLong <owen at delong.com> wrote:
>> BCP38 was out before RIRs started moving down to /24s. It didn’t take long
>> for ISPs to adapt to /24s from /20s.
> 
> Hi Owen,
> 
> Most ISPs never adopted /20 in the first place. The were /24 before
> and after. Those that adopted /20 were mostly aware of what goes on
> here so had only the trouble they made for themselves.

Actually, many adopted /20 at the time. At the time, their routers were also
close to collapse. Over time, things got better (the growth in prefixes slowed
long enough for technology to get ahead).

> /24 is different. Announcements smaller than /24 were interdicted
> right at the start of CIDR by people who very nearly been burned by
> too many announcements. The /24 filter is very widespread.

Yes, it is. I have no illusions that the /28 filter will be immediately widespread
for this prefix. However, it doesn’t necessarily need to be. If it gets adopted
by a select handful of the providers that tend to be the most aware of what
is happening here, then so long as the more specifics make it from the
end-site in question to one of those providers, chances are pretty good
that it will be generally workable.

Is it ideal? No.

However, I will point out that anything involving IPv4 at this point is far from
ideal.

> I have no objection to changing the size to a flat /24. I do wonder
> about the efficacy of the policy overall. For something this
> transitory, is there a sound reason why addresses assigned from an
> upstream ISP are not suitable?
> 
> What is ARIN's experience with allocations under this policy? Have
> there been any? What were the justifications?

I don’t anticipate that there will be any desire or need for significant
allocations under this policy until after some ISPs are no longer able
to issue these longer prefixes to their customers. This is intended to
provide a set-aside for new entrants after that point in case there are
enough legacy holdouts that have not adopted IPv6 as to make it
impossible to be “on the internet” without being at least able to talk
to the IPv4 internet.

I think there have been a small number of applications under this policy,
but I don’t know for sure. However, I suspect them to be more a matter
of attempts at creative application of policy than the core need for which
the policy was intended which does not exist as yet.

Owen