Re: [sig-policy] IPv6 proposals summary and call for discussion
On Feb 8, 2011, at 11:07 AM, Andy Linton wrote:
> I think prop-083 and prop-087 are in many ways talking about the same
> thing. Our current view of network size is still influenced strongly by
> an IPv4 of the world and we see a /32 as so huge that for most entities
> they'll never need anything else. And to a large degree that's true.
>
They aren't actually.
Prop 083 is about providers that have multiple locations that aren't
connected by an interior backbone.
Prop 087 is about technologies like 6rd.
Neither one of them has much to do with the /32 perception problem
that prop 090 attempts to address.
> But there are situations where people will need more than one /32 and
> that should be possible. So I'd like to see us come up with a policy
> that made it clear that larger or multiple delegations of address space
> can be made for clear, well documented, technically sound reasons.
>
> We could then have the cases that these two proposals cite listed in an
> Appendix to the policy. New valid reasons could be added if/when needed.
>
I wouldn't have a problem with that approach.
> The policy should enshrine the principles only.
>
> So if we could come up with a way to unify these proposals I'd be happy
> to support that but I won't reject these if we can't manage that.
>
> Looking at prop-090 I have no issue with the change in calculation. I
> don't believe there's a "correct solution" to this. I note that the
> percentages quoted in Section 2.1 and Section 5 don't agree with each
> other but if either of the sets of numbers is in the right ball park
> then it doesn't matter.
>
Good catch... That's a typo in section 5. 99.54% is the correct number.
> I like the idea of the nibble alignment in Section 4.3.
>
> I'm less keen on the allocation criteria in Section 4.5. Sections 4.5.2
> and 4.5.3 as written prevent any new organisation who isn't an ISP from
> obtaining IPv6 address space for their own use.
>
That's certainly not the intent. The intent is not to change the end-user
policy as it exists today and only to modify the LIR/ISP policy. This
may be an unintended side effect of my limited familiarity with the
APNIC policy environment.
Can you suggest text that would rectify this issue?
> So a new university, government department or company who may have many
> hundreds or thousands of users can't multihome using IPv6 because they
> don't plan to give out address space to other organisations?
>
> We had criteria like these in place for IPv4 because we've recognised
> for many years it was a scarce resource and people have fabricated
> requests to the RIRs to justify their wants. Let's have delegation
> criteria but not these ones.
>
Agreed.
> I'm also doubtful about Section 4.8. Any policy that says a recipient
> needs to renumber within 5 years will never work in practice. What
> happens if they don't renumber? What sanctions will APNIC be able to apply?
>
I expect the number of organizations affected by this to be relatively small.
At the end of 5 years, APNIC can revoke the original block, terminate
its ip6.arpa records and whois entries. Eventually, if APNIC so chose, they
could allocate the block to another requestor.
Since this trade-in policy is intended only for LIR/ISP organizations that
are very early in their rollout and want to start over with the larger block,
returning the original smaller block shouldn't be a sticking point.
On the other hand, if they're actually using a significant portion of
the smaller block, then, they might qualify for subsequent allocation
under the other sections of this policy which do not require return.
Does that address your concern?
Owen