[arin-ppml] Encouraging IPv6 Transition (was: Clarify /29 assignment identification requirement)

[Chris Grundemann]

On Sun, May 13, 2012 at 3:53 PM, Jimmy Hess <mysidia at gmail.com> wrote:

Thanks for the reply Jimmy, some good points indeed - my responses are
inline, below.

> On 5/13/12, Chris Grundemann <cgrundemann at gmail.com> wrote:

>> It's not a huge win because it won't make any appreciable difference,
>> it's not actually a win at all.
>
> It removes a barrier in IPv6 deployment;  which is not just about
> cost, but also about time
> and paperwork.

Full agreement. Time and paperwork = cost; cost is a barrier...

> It also greatly simplifies  the allocation of
> addresses,  and therefore eliminates address space management overhead
> expenses for the RIR.
> Carve one /14  to represent /48 end-user assignments  a priori based
> on telephone number.
> Carve one /8 to represent  a IPv6 /24 end-user assignment  for each
> 16-bit AS number.
>
> Any allocation made in an unstructured way is then a "special allocation"

Forcing folks who are unwilling to ask for IPv6 to decipher an
algorithm, however simple the algorithm, is not simpler (for them)
than assigning them a specific block of addresses directly (when they
ask or them). Again, if they lack the motivation to apply for IPv6,
what can lead us to believe that they will sit down and hash out their
automatic assignment (of which they are very likely to be unaware)?

>> Let's ask the real question here: How does blindly assigning IPv6 to
>> organizations provide any additional motivation to deploy IPv6? If
>
> It takes away and excuse not to deploy IPv6.

So noted. To make my stance more clear: If the only thing standing
between orgX and IPv6 deployment is the cost of applying to ARIN for
IPv6 space (time, paperwork and fees), then I have no confidence that
automagically assigning space to orgX will actually cause them to
deploy it. Yes, it makes it (marginally) easier to do so, but without
any organizational motivation, it still will not happen.

>> We could also revisit the formerly unpopular idea of requiring IPv6
>> deployment in order to obtain additional IPv4 addresses...
>
> Added barriers to deployment of IPv4  without IPv6 is also a possibility.
> But based on your argument.... just obtaining the addresses doesn't
> mean they will deploy IPV6.
>
> They may simply obtain IPv6 addresses,  without deploying them,  so that their
> burden of obtaining more IPv4 resources is reduced.
>
>
> Perhaps we should have a  3 month  SUPPLY  rule on both allocations
> and transfers
> for organizations  that have not demonstrated IPV6 deployment,
>
> A 12 months of IPv4 resources supply rule on allocations and transfers
> to organizations
> that have demonstrated at least 30%  IPv6 deployment.

Noting that this has been unpopular in the past, and that I (for the
most part) agree with Mr. Dart that the implications of this may very
well overstep our mandate of stewardship; there are many ways to
determine deployment of IPv6 which are more substantial than simply
'we have received IPv6 addresses from an RIR.' Route announcement, DNS
entries, IPv6 traffic levels, SWIP/WHOIS, configuration files, etc...
This is absolutely do-able in a meaningful way, I think the question
is rather whether or not the community feels comfortable doing it.

Cheers,
~Chris

>
> --
> -JH

-- 
@ChrisGrundemann
http://chrisgrundemann.com