[ppml] 2005-1 or its logical successor

[Lea Roberts]

hi Bill (et al) -

as you might have guessed from my comments in the hall,
I don't quite agree with you on this...

On Thu, 27 Oct 2005, Bill Woodcock wrote:

>       On Thu, 27 Oct 2005, Marshall Eubanks wrote:
>     > For those who couldn't make it  to L.A. (I am assuming  you did), could you provide some
>     > context for this ?
>
> A revised-by-committee manglement of 2005-1 was presented, which had a ton
> of restrictions piled on top of it, which made it useless.  That is, there
> was no remaining constituency for it.  A huge queue of speakers denounced
> it, and only one person spoke in its favor, Dan Golding, and he said the
> only reason he supported it was because he thought fixing it later would
> be faster than revising it again.  Fortunately, sanity prevailed, and the
> whole thing was sent back to the drawing board to have all the junk
> stripped back off.

for the record, the straw poll was 24 vs 25.
>
> Which, I think, brings us back to the original 2005-1.
>
> So my question was whether there's anyone who, having seen the much
> nastier alternative, doesn't now support the original 2005-1?
>

I understand that it would be really nice if we could really reach
consensus to not care about the size of the IPv6 routing table, but that
wasn't the reality I experienced in Orlando (where the split in the room
was about even as well).  Routing table size was a major concern sited by
those whose indicated opposition to the original 2005-1.

For the record, here are some excerpts from what I wrote to Kevin Loch
inviting him to work on the combined 2005-1 that you are so unhappy about.
I hope this will help others (and maybe even you) to understand how the
extra stuff got piled on.  I honestly believe that for 2005-1 to achieve
consensus it has to focus on the large organizations for whom renumbering
is a problem while having criteria which would limit the number of
qualifying organizations to the range of 5K to 10K.  When I wrote the text
below, the feedback from other AC members is that it matched their
recollection from ARIN XV.

I gather you don't agree.  Hopefully we'll hear more from others on
PPML...  :-)
				sincerely,	/Lea Roberts, ARIN AC

as written on 31-August-2005:

Hi Kevin -

Earlier today, the ARIN Advisory Council had a teleconference to act
on the recently submitted policy proposals.  By a close vote, a motion
to move your proposal, "IPv6 Direct assignments to end sites", forward
as submitted was defeated.  The AC members present then unanimously
voted for the option to work with you to see if it would be possible
to combine your proposal with the existing 2005-1 policy proposal,
which, although it failed to reach consensus in the previous ARIN
meeting, the AC believed had enough support to work with the author to
craft a proposal which might reach consensus.  As the AC member who
has been working with Owen DeLong on 2005-1, I volunteered to be the
one to contact you regarding this choice.  As you and Owen have
already stated on PPML, both proposals aim in a very similar direction
and so it should be easy to combine them.  However, before you agree,
I would like to inform you about the direction that the changes to
2005-1 are taking as they may not be to your liking.

I have been working with Owen to modify policy proposal 2005-1 in
response to the concerns of those who opposed it.  I don't know if you
were at the meeting, but there were a number of objections to the
original 2005-1, the main ones being a concern over a run on AS
numbers, which are currently the most constrained Internet Resource
until 4-byte ASN's are a reality, and major concerns over the
possibility of a large increase in the size of the IPv6 default-free
routing table.  There were assertions that it was too early for making
multi-homing a rationale for a direct assignment of IPv6 address
space, unless it was only for a limited time, until the viability of
the shim6 effort in IETF could be determined.  While the current
number of sites who multi-home could easily be accomodated at this
time, the effect of an IPv6 policy has to be looked at over the
multiple 10s of years that IPv6 will need to be functional.  Very few
people believed that limited time assignments were viable (i.e. could
actually be reclaimed) and asserted that it would create a similar
situation to IPv4, where early adopters have an unfair advantage.  In
support of the proposal, a number of commercial companies, who were
attending the co-located NAv6TF meeting, expressed their unwillingness
to invest resources in deploying IPv6 with Provider Assigned address
space, as they were unwilling to be "locked in" to a provider or else
have to renumber their entire enterprise.  When the sense of the room
was taken, the attendees were about evenly split and so there was
clearly not a consensus.

I have been relying, as much if not more, on hallway conversations
with those who opposed the advancement of 2005-1 to try to craft a new
version of the proposal which could achieve consensus sooner rather
than later.  Most of those who opposed 2005-1 admitted that the design
concept of only Provider Assigned space for IPv6 was clearly no longer
tenable, but they were very concerned about almost unrestricted access
to Provider Independent IPv6 address assignments.  They indicated that
it was too early in the protocol's lifetime to allow unrestricted
routing table growth and expressed the hope that shim6 might still be
successful.  There is a real belief that IPv4-like multi-homing will
doom the IPv6 routing table to grow beyond a workable size and some
other solution has to be found!  They expressed an understanding of
the large enterprise renumbering problem and indicated that they would
support a policy that provided for PI address assignments to a small
number of large organizations for whom the cost of renumbering would
be a significant expense.  FYI, I think that once a PI assignment
policy exists, the requirements can continue to change over time, much
like the current IPv4 assignment size continues to become less
restrictive.  I think that getting over the "only Provider Assigned"
hump would be a major policy achievement that's within sight.