[arin-ppml] Policy Proposal 93: Predicable IPv4 Run Out by Prefix Size - Revised

[Tom Vest]

On Jun 19, 2009, at 11:13 AM, <michael.dillon at bt.com> wrote:

>
>>>> At a bare minimum, new entrants must be safeguarded.
>>>
>>> Not in the end-game.
>>
>> How do you justify this assertion?
>> Do you think that accepting industry closure to new entry is
>> a good idea in general?
>
> The end-game is created by circumstance and it is the circumstance
> that closes the industry to new entrants. Just like oil production
> in Pennsylvania.

That would be a good analogy IF Pennsylvania oil producers controlled  
the only oil on Earth, AND Pennsylvania oil producers could continue  
to service their own customers indefinitely regardless of the size of  
the oil reserves, AND they also unilaterally controlled the  
feasibility/cost of shifting to a different energy source for every  
other current or future energy user.

In other words, it's not a very good analogy.

> There will eventually be a new IPv4 game once the Internet game has
> shifted onto IPv6, and at that time, new entrants will once again
> be able to build IPv4 networks, again because of circumstances.
> Just like repurposing a steel mill from processing ore to processing
> scrap metal.

No one really cares about about building "IPv4 networks" -- not today,  
nor in all likelihood anytime in the future.
They care about building IP networks that work, which includes the  
feature "can interoperate independently with more than 0.1% of the  
rest of the Internet."
Today that happens to limit the options to IPv4-based networks.
If the environment eventually changes enough to make IPv6 functional  
enough to stand in for IPv4, then the ambition will still be the same;  
only the implementation options will change.

The goal is to preserve industry openness as long as it takes for that  
environmental shift to occur.

>> Do you think that the current balance of internal (i.e., current
>> member) and external stakeholder interests, which favors the
>> existing industry coordination arrangements, will survive
>> that development?
>
> Huh?
> This is the PUBLIC policy mailing list. Please write in English.

Every governance arrangement that endures for any length of time does  
so because of a combination of internal and external support.
One of the best ways to preserve that support, on both sides, is to  
keep the barrier between internal and external stakeholders low, so  
it's easy to transition back and forth as circumstances dictate. One  
of the fastest ways to erode that support is to let the barrier  
ossify, creating a permanent caste of privileged insiders vs. everyone  
else.

Hopefully that's clear enough.

>>> In any case, there is plenty of IPv6 for new entrants so ARIN is
>>> safeguarding them.
>>
>> I'm glad that IANA didn't think that way thirty years ago:
>> "256 /8 recipients could easily support plenty of future
>> customers, so the system is safeguarding them."
>
> But IANA did think that way more or less.

The "less" part of that thinking is why there are now around 15-20k  
independent players in the routing services industry, instead of just  
256.
I personally don't think that that's a trivial difference.

> But when circumstances changed, the thinking changed as well.

Agreed, but in the previous two episodes, the thinking changed *in  
anticipation* of the changing circumstances, and the pro-competitive  
adjustments in industry practices that resulted *averted* the worst of  
the anticipated consequences. I wish that model of industry adaptation  
were more widely appreciated today.

>> The fiction that IPv6 is substitutable for IPv4 *today* will
>> not hold; it will never hold until it ceases to be fiction.
>> The timing of that development will be wholly determined by
>> incumbent IPv4 holders.
>
> The fact is that the global IPv4 network will not be able to grow
> in a few years and after two years, growth will be harder and more
> costly. But IPv6 can grow today, two years from now and twenty years
> from now.

Does that mean that you're willing to start returning all of your IPv4  
to the registry today, in return for IPv6?

> This is fact, not fiction. The only reason that IPv6 and
> IPv4 are not easily substitutable today is because the networking
> industry (equipment vendors and operators) are dragging their feet.

But Michael, when we say "the industry," we're talking about us.
There is no "they" in "dragging their feet" that is not us.
It's a description of the fact, not an explanation -- much less an  
explanation that instills confidence that the problem will simply  
disappear by itself anytime soon.

> The OS industry has been ready for years now and people who turn on
> IPv6 find that it just works, until they bang their heads against
> network operators who in turn are banging their heads against
> various vendors. Even Cisco and Juniper, who are the best of the
> lot concerning IPv6, still have some rough spots.
>
>> How many investors were still funding real estate speculation
>> in Q4 2007?
>
> You convinced me. ARIN should help new entrants to throw away
> their money on new IPv4 networks right up until the last day.
> Let the bankruptcy courts resolve the issue.

The only thing that ARIN and/or industry members can do, at any/every  
point in time, is take steps to make sure that some kind of  
*functional* IP addresses remain available to aspiring and eligible  
new entrants on a non-adversarial, non-discriminatory basis -- kind of  
like they way they are today. Availability of non-functional IP  
addresses, or availability of functional addresses but only under  
crippling conditions are not real alternatives; they're not going to  
fool anybody for long.

If there's nothing that ARIN and/or industry members can do to  
preserve that condition, then the future is predetermined, and I guess  
we can all just relax.

>> The chaos and "dying game" that you're describing will be
>> internal to current IPv4-based operators, who need to decide
>> out what to do next.
>
> No, it will affect anyone running an IPv4 network that needs
> to grow. That includes end users and new entrants who are not
> currently running IPv4 networks.

An incumbent IPv4 holder who needs to grow in a post-IPv4 environment  
has three choices:

1. Beg/borrow/buy IPv4 from another incumbent
2. RFC 1918 + reshuffle
3. IPv6

A new entrant will have, at best, (1) -- but they'll be competing with  
all of the incumbents who will still be preferring that over the  
alternatives.
The winning public argument for transfer markets depended heavily on  
the assertion that there would be substantial continuing demand for  
IPv4 *among incumbents.*
Without that element, the transfer proposals would have taken on an  
entirely different character.

>> The most visible/prominent manifestation that the runout will
>> have in the broader context will be the stories of what
>> happens to the next generation of aspiring operators who find
>> out what our version of "safeguarding them" really means.
>
> Nah, the press likes an underdog. They will be writing about the
> farsighting new entrants whose IPv6 network is running rings around
> the incumbents who are struggling to get IPv6 product out of the lab.

Or perhaps they'll be writing about the clever operators using IPv10,  
which will be just as useful to new entrants in an IPv4-mandatory  
market.

But I think that neither is likely.

When read the papers and magazines "of record" these days, I don't see  
a lot of articles about scrappy new banks running rings around their  
stodgy, misguided predecessors.
I do see lots of articles about banks -- but most of them have a  
decidedly different story line.

nuff said,

TV


> --Michael Dillon
>
> P.S. that bit about helping new entrants throw away money was
> sarcasm, in case anyone thought I was serious.

I got that ;-)
For anyone who didn't, please note that I was being similarly glib  
with the "IPv10" reference...