[arin-ppml] Policy Proposal: Dedicated IPv4 block to facilitate IPv6 deployment

On Jun 7, 2008, at 7:13 PM, Paul Vixie wrote:

>>> ...  if the AC can design and accept experiments that would inform  
>>> this
>>> policy, i feel sure that we could get the experiments run by various
>>> nonpartisans with expertise in the area.
>>>
>>> paul
>>
>>   Renesys did a lightning talk on this topic at NANOG 41 --
>> <http://www.nanog.org/mtg-0710/presentations/renesys- 
>> lighting.pdf>.  They
>> have a page with a few provider's policies.  Both AT&T and Level3  
>> have a /24
>> limit.
>>
>> -Larry
>
> do we expect these policies to remain in force, or loosen, or  
> tighten, after
> iana runout occurs and there is pressure to further deaggregate the  
> ipv4 space
> and fill in all these little holes?  (and if so, why, or if not, why  
> not?)
>
I would expect that initially, they will loosen, but, then as it becomes
harder to deal with the tide of increasing route table entries, they
will again tighten. I think there will continue to be various forms of
economic and technical pressures that create a sort of shove-of-
war (hard to call it a tug in this case as all push and no pull).

I think that situation will persist and potentially become worse
until we reach the point where people start to remove their IPv4
deployments. I don't expect that to be any time soon.

> i think that the industry's early reaction to iana runout and/or RIR  
> runout
> will be to loosen filters since a lot of address blocks and  
> therefore a lot
> of potential customers would be hidden by the filters described  
> above (/24)
> and the initial desire will be to continue growing since a lot of  
> the big guys
> have megaroute cores.  competition of the form "i'll charge you less  
> but i'll
> have to NAT you" won't always win vs. competition of the form "i'll  
> charge
> you the same, or i'll charge you more, but i won't NAT you."   
> smaller ISP's
> without megaroute cores won't have good leverage if they want to  
> ignore these
> routes.  hierarchical routing, as a principle, will take an arrow in  
> the neck,
> and won't return until non-hierarchical growth proves measureably  
> impossible.
>
I'd say that's about the same as my prediction.  The one potential
difference being that I think the cycle to get to where restrictions are
increasing will only be 6-12 months after the first big RIR exhaustion.

Owen