[ppml] NANOG IPv4 Exhaustion BoF

[Tom Vest]

On Mar 6, 2008, at 6:23 PM, Geoff Huston wrote:

> Tom Vest wrote:
>
>>>
>>> And drive the industry to IPv6 deployment.
>>>
>>> Geoff
>> I think we all wish this is how things will go.
>> But it could just as easily go the other way, and either vector  
>> could easily be pre-empted by major unwelcome surprises.
>> Randy's very educational demonstrations have revealed that there's  
>> a long way to go before anything is 100% certain.
>> In the mean time, new entrants are not going to be the ones to  
>> pioneer this effort, because they'd still have to wait for the big  
>> guys to get their act together in order to be able to benefit much  
>> from entering the business. So if anyone is to lead, it will be  
>> the big guys. Do we believe that they're holding out until they  
>> know they can monetize their future IPv4 surplus assets?  Do we  
>> believe that they're simply unable to afford the transition unless  
>> they can extract large sums from the IPv4 trade? Who is going to  
>> provide those sums, if not other big operators? Or do we think  
>> that adding a seven digit number to the cost of starting up will  
>> fill the gap without sparking internal rebellions and external  
>> interventions?
>> One could also say that OPEC and increasing "real" oil scarcity is  
>> leading to the adoption of hydrogen fuel cell technology. In fact  
>> one could have said that back in the early 1970s too; we could  
>> still be saying in 20-30 years from now. Burning bridges behind  
>> you when the future is still uncertain just doesn't seem like  
>> prudent planning to me...
>
> About the only time we had a choice in this particular saga was  
> around 4 - 5 years back when the post-bust economic recovery lifted  
> the Ipv4 address consumption rate from around 4 /8's per year to  
> around 10 /8's per year. At that time there was still sufficient  
> space in the unallocated address pool to support a dual stack  
> transition without undue escalation of the pain barrier. But there  
> were no economic signals whereby this could be expressed. The cost  
> of IPv4 deployments continued to fall in unit cost terms so its  
> little wonder than there was no serious deployment of IPv6 - the  
> underlying differential pricing signals were simply not present and  
> IPv6 was additional cost without any incremental revenue  
> opportunities.
>
> But, like it or not Tom, the bridge behind us is already in cinders  
> (I suppose I should be thankful that the analogy has moved on from   
> sinking ships, stampedes, and train crashes to burning  
> bridges ! :-) ). However, I find it hard to sift through your words  
> here to extract what you are simply stating as a preferred outcome,  
> so instead let me restate my perspective here.

I believe that there are still credible alternatives to a  
decentralized market that could do better on all of the important  
short-term fronts: sustain some level of IPv4 liquidity (mostly for  
new entrants, 6/4 gateways, xTRs, etc.), keep the industry open to  
new entrants, preserve needs-based allocation principles, preserve  
the central registry function, resist internal manipulation and  
external intervention, etc. -- to basically do all of the things  
necessary to sustain an orderly, incremental, self-governed (i.e.,  
"successful") transition. For me that is the only goal that matters.

> Scarcity in IPv4 addresses will be reflected in price in any market  
> scenario.

Yes. But your statement is a restricted form of the incontrovertible  
general fact: scarcity will affect availability in any/every kind of  
system for allocating values, including "markets".

> When demand outstrips supply this is an entirely conventional outcome.

Markets are not inevitable. Demand for IPv4 outstripped "sustainable  
supply" at almost every moment since it was invented. Why weren't  
markets for Class B space inevitable back the last time we visited  
this territory?

> The market also has characteristics that lead to expectations of  
> high volatility. Again, this is not uncommon and frankly really  
> cannot be avoided.

If markets lead to high volatility, but other allocation systems may  
not, then does it make sense to just accept volatility as inevitable?  
Who benefits from volatility? If markets also lead to expectations of  
loss of other important collateral values (e.g., openness, industry  
solidarity/competence, independence, etc.), who benefits from that?

> However, there is a form of substitution that in and of itself is  
> not without cost, but when the price of IPv4 in such a market  
> exceeds the substitution cost of deployment of IPv6 then the market  
> signal regarding IPv6 deployment would be clear to all.

My concern is that there is a very high probability that the  
"critical infrastructure self-governance failure" signal will be much  
clearer to a much broader audience long before the other kind of  
signal percolates up and down far enough.

> So one can either say "the future is this big scary unknown place  
> that we shouldn't tamper with", or you can do what you can to  
> mitigate some of the more destructive potential outcomes and  
> attempt to encourage some of the more beneficial ones.

Oh I agree 100%, and am acting on that belief, just as I believe you  
are. We just seem to be scared and reassured by different things.

> Now if your aim is to make IPv4 last forever then obviously we  
> disagree about what is a net beneficial outcome for this Internet.

Come on Geoff ;-)
I am not going to tax the patience/credulity of the audience by  
insinuating that you have some silly or sinister ulterior motive.
But I will look forward to continuing the discussion, and hopefully  
contribute to some set of policies that we can have (some)(more)  
confidence in.

TV