[ppml] 2046 was Re: Summary of Trial Balloons

[Edward Lewis]

At 16:34 -0500 3/30/07, Jim Weyand wrote:

>It seems like it is time to start the relatively hard work of actually
>developing alternative policy proposals to deal with the IPv4 Address
>Exhaustion Issue.

I'm following up on this message just as an anchor point, having read 
the large volume of mail on this.

I'll start with the negative statements as I don't have a divine 
answer myself and I have doubts on many of the proposals to date.

I don't think that ARIN has as a goal of promoting IPv6 over IPv4. 
This isn't about one technology or the other.  I think that the 
"battle" over that has to happen in protocol and applications arenas 
and ultimately in the marketplace.  But I still think it is important 
that ARIN help wean us off of IPv4.

I don't really care about address reclamation procedures as answers 
to the problem.  Although I am all for efficient use of the resource 
and for ARIN taking as active a role as permitted in doing this, IPv4 
ultimately will not support a truly global Internet - not in my 
opinion.  Squeezing a few more years only postpones the inevitable.

I don't think that ARIN can really use any economic tools (I won't 
say "fees" to keep Lee happy) to make IPv6 adoption happen - whether 
it is "taxing" IPv4 or giving/assigning away IPv6 for free.  The 
reason is that the cost of moving to IPv6 is not just the cost of IP 
service, there are other costs involved.

My motivation for caring about this issue is that a future in which 
we have a segmented network, whether it is IPv4 in North America and 
IPv6 in Asia or a plethora of NAT boxes or 6-to-4 boxes gluing it all 
together, is not desirable.

We are already saddled with "legacy IP resource assignment" - the 
swamp, the pre-RIR assigned space with unclear policy and 
"ownership."  Do we also want to see a "black-market" if you will of 
IP addressing resources?  Wouldn't that create an economic drag on 
this system?

The are a lot of hypothetical futures already presented and we can 
argue over who's more right but I think that would be futile.  We can 
also argue over a lot of other details, such as how big a prefix an 
ISP should allocate to an end user and so on. Ultimately though we 
(on this list) need to figure out what ARIN policy for address 
registration is needed (and perhaps that answer is to do nothing).

>8) An informal proposal to get endusers to demand access to IPv6 networks
>by creating a media storm similar to Y2K.

Of all of the trial balloons that are listed, this seems the oddest 
but also the most alluring.  I don't like the idea of a media storm 
because that's all bluff and bluster.  But there is a parallel to the 
Y2K situation - or maybe the year that 32 bit clocks roll over to "0" 
again, which is what 2036-2038?  I've forgotten.  The parallel is 
that we are running towards a point in history when a one-time 
engineering assumption is about to be broken - that 32 bits were 
enough for an address.  It's like explaining to someone that "1984" 
wasn't a year, it is a futuristic novel.

Perhaps the greatest benefit ARIN is providing here is giving us a 
forum to discuss the issue.  What I am thinking of is the post from 
Jason Schiller giving details on the problems an ISP faces in 
providing IPv6 service.  We have a few opinions on what holds 
enterprises back, what motivates home consumers, and my perspective 
is from someone that has a service that is or ought to be provided 
over IPv6 in addition to IPv4 (as long as they are the two choices).

I think that the bottlenecks to IPv6 adoption have already been 
presented on this list.  I am not clear on what is the easiest one to 
remove nor the most important one to remove.  I am not sure that a 
media barrage about IP6 would help because it might actually hurt our 
adoption of IPv6 - a vendor is demanding a large license fee for the 
software that does IPv6 on one of our "front matter" boxes.  ("You 
need this new technology right? So you will pay more!")

At the APNIC meeting, the least appealing part of ARIN's 2007-12 (12! 
It's only April) was the termination date.  BTW, here are excerpted 
notes from APNIC on the discussion:

http://www.apnic.net/mailing-lists/sig-policy/archive/2007/03/msg00015.html

#Discussion at APNIC 23
#----------------------
#
#There was general support the following three principals in the
#proposal:
#
#    - Global synchronization
#    - Keeping current practices until the last moment
#    - Recovery of unused address space should be discussed
#      separately
#
#The remaining principle, "Some Blocks to be left", was split into
#two:
#
#    - Need to define the last date of allocation in advance
#    - Keep some blocks reserved after the defined last date of
#      allocation
#
#There was no consensus on these two elements of the remaining
#principle.

I think that the first three principles are okay and I don't see the 
need really to keep some reserve.  But I am more receptive to a 
termination date of the use of IPv4 - note I said "use" and not 
"allocation."

I know, it's ludicrous, it's crazy.  But maybe if we state that in 
the year 2046 IPv4 will be no longer recognized on the public 
Internet there will be just the right movement to accomplish - well - 
whatever "we" want.  Of course, that's not something ARIN does - but 
maybe a global policy of retiring the IPv4 number resource by 2046 
can be carried to IANA?
-- 
-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
Edward Lewis                                                +1-571-434-5468
NeuStar

Sarcasm doesn't scale.