[arin-ppml] Q1 - ARIN address transfer policy: why the trigger date?

[Owen DeLong]

On Jun 22, 2008, at 11:07 AM, Joe Maimon wrote:

>
>
> Owen DeLong wrote:
>> On Jun 20, 2008, at 10:10 AM, Paul Vixie wrote:
>>>>> isn't this inevitable in times of shortage, ...
>>>> I don't accept your premise that a market is inevitable.
>
> <snip>
>
>> animal to extinction and then the market collapsed.  In this case,
>> we have the option of just watching the extinction occur without
>> creating a market
>
> If there will be a market, it will happen with or without the  
> registries.
>
Probably, but, without the registries, I expect the market to be  
significantly
smaller and less widely accepted.  I also expect it to be less  
transparent
and riskier to the market players.

> They will get their address space one way or another, and they will  
> demand it be routed. The ones paying the paychecks will demand they  
> get routed. Those who refuse to route it will cease getting paychecks.
>
I don't see how this will happen.  I see lots of people insisting on  
this position,
but, the reality is that you may redistribute that small unused  
portion of the
space which exists today, but, once that's gone, I don't see the  
continuing
market scenario being all that effective or likely, even with the  
registries.

Also, while I think they will get routed by their direct upstreams, I  
think you'll
see increasing fragmentation of routing as people not getting paid to  
route
a particular prefix begin to filter the smaller blocks they aren't  
getting paid
for out of necessity.  Eventually, that will essentially eliminate the  
smaller
blocks from being at all useful.  I don't see that as being a win for  
the
community.

> Effectively at that point ARIN is increasingly irrelevant. We dont  
> want this scenario.
>
I disagree.  I think that is the point where ARIN preserving some  
level of
addressing integrity by not allowing random transfers will be the only  
path
to possible recovery of the routing domain for IPv4, and, even then, I  
am
not so sure it will work.

> This is already happening, according to anecdotal evidence.
>
In a very limited scope and with tremendous risk to the players
involved.  Armed robbery is already happening in most cities
in America.  That does not mean I want the government to start
selling armed robbery permits.

> The only way this scenario doesnt play out this way is if the PHB's  
> say, "sorry, according to my engineers, no ipv4, only ipv6", and the  
> customers say "Great! I didnt actually want the ipv4 anyways".
>
I think the reality will be somewhere between those two extremes
and will slowly work its way closer to the "no ipv4, only ipv6"
extreme as you described it.

> Depending on who you ask, thats either possible, impossible,  
> unlikely, likely, unrealistic, naive, idealistic, overly simplistic  
> or something or another. We dont actually know how fast things will  
> transition to the point ipv4 has no value and ipv6 has all value.
>
It's overly simplistic in my opinion.  There's a lot of grey between  
those
versions of black and white and the reality will be in many different
shades of gray for many different subsets of ISPs.

> Thats the only way there wont be a market in ipv4. If it has no  
> market value.
>
I'm not saying we won't have a market.  I'm saying that in my view of
things as they have played out so far, legitimizing the market does
not provide much benefit and definitely increases risk.

> So if there will be a market, it should be one where we can make it  
> attractive for those we want to participate, to participate.
>

> In exchange we can gain by imposing a cost of participation inline  
> with our needs.

> The balance that needs to be struck is between cost of participation  
> and the attractiveness of participation.
>
So, by that logic, since there will be armed robbery, the government  
should
simply accept that and sell armed robbery permits to benefit from the  
revenue
that is possible and do so on a basis that allows us to select more  
desirable
armed robbers?

>
> And if there wont be a market, its unlikely that ARIN could create  
> one.
>
There will be a market.  The true question is whether legitimizing and
expanding it is a good idea.  From everything I have seen, legitimizing
and expanding the market carries a lot of risk both to IPv4 and to IPv6
future policies with very little reward potential.

> So coming to grips with a potential increase of value IPv4 blocks  
> will have post-runout and developing an approach for it are all  
> about remaining relevant, so as to attempt to ensure our needs are  
> met.
>
I think ARIN's best way to remain relevant is to continue as we have
to be good stewards of the existing IP free-pools and IP policy in
general.  In the future, that's going to be IPv6.  For now, there is
a dwindling role in IPv4.  Extending that role by legitimizing a
market and expanding it seems to me to be a very short-term
extension and a very limited potential benefit with a whole lot of
potential problem areas.

> A relevant ARIN is in all our best interests at this point.
>
Sure.  I don't think that ARIN will become irrelevant just because
there is a market in IPv4 addresses that ARIN is not involved in.
I think that IPv6 will become increasingly relevant and I don't
see a black market in IPv6 replacing ARIN.  I also don't see
people developing an alternate ASN registry as being a particularly
likely scenario.

Owen