On Jun 25 15:29, William Herrin wrote:
> The question for this thread is: What limits, if any, should be placed
> on an organization's ability to get IPv6 addresses from ARIN?
>
> Should they have to be for use on the Internet with off-Internet uses
> expected to use something like ULA? Or are off-Internet uses just as
> qualified as Internet uses?
I'd like to see ULAs go the way of the dodo in enterprise environments.
They have caused me nothing but problems, in part because of glibc's
address selection algorithm. I believe that there are good technical
reasons for GUAs to be used off-Internet, and provided an organization
does not intend to use the *majority* of its addresses off-internet, it
should be an allowable justification.
> Should they have to follow an addressing plan which supports
> hierarchical aggregation? Or is that a network engineering question
> into which ARIN should not pry?
None of ARIN's business. We've messed it up more than once. Best
practices are going to evolve more in the future.
> Should sub-sub-allocations be restricted (i.e. if an external org
> needs to allocate addresses to yet another org, they have to get their
> IP addresses from ARIN instead of from you)? For example, NRPM
> 6.5.2.1e says that any ISP customer who needs more addresses than a
> /32 must get their IPv6 addresses directly from ARIN.
As long as suballocations and reallocations are properly documented in
RDAP/rwhois, tooling supports RPKI and IRR registrations for
reallocated orgs, and so forth, I think there is no reason to restrict
it unless it is obviously intended to skirt or abuse ARIN policies.
However, because of RPKI alone, we want to discourage it: most users
will be happier with a direct contract with ARIN.
> Should minimum allocations be set based on the registrant's IPv4
> holdings? For example, NRPM 6.5.2.1g says ISPs are not allowed to have
> a longer prefix than /36 unless they have no more than a single IPv4
> /24.
This rule should be removed. If someone wants a smaller allocation to
engineer around the fee schedule, they should be allowed to do that.
> Are there any other advisabile limits on who should be able to get
> IPv6 addresses? Any existing limits that should be changed or done
> away with? Your views are respectfully requested.
The maximum initial LIR allocation is a good idea, though we may need to
revisit that size and the text. I think the "reasonable technical
justification" rule for PI space is appropriate across the board. If we
retain any formulae for identifying allocation size, it should result in
a range of justifiable prefix lengths: e.g. for a given number of sites
and customers you are eligible for a /36 to a /28: pick what you want
(since you have to pay the bill and engineer the network). Obviously we
don't want to build the NRPM around the fee schedule, but for some
organizations that smaller fee is worth it.
Eric C. Landgraf
Virginia Tech
_______________________________________________
ARIN-PPML
You are receiving this message because you are subscribed to
the ARIN Public Policy Mailing List ([email protected]).
Unsubscribe or manage your mailing list subscription at:
https://lists.arin.net/mailman/listinfo/arin-ppml
Please contact [email protected] if you experience any issues.
