[arin-ppml] IPv4 Depletion as an ARIN policy concern

[Lee Howard]

> > Sounds like there may be some dollars to be waived under the noses
> > of some transit providers.  "Give me production-grade IPv6 transit
> > or my IPv6 bits go elsewhere."
> 
> Lee,
> 
> I assure you it only sounds that way. The investment is a cumulative
> effect measured not just in terms of my transit connection but
> simultaneously in terms of yous and the transit for every other
> individual who would use IPv6 to talk to me. I could wave Bill Gates'
> fortune at the transit providers and it would still be years before
> the IPv4 and IPv6 DFZs' standard of quality reached parity.

I understand and agree that the "network effect" is at work here.
But I hope that transit providers on this list are noticing that reliability
of IPv6 connectivity is high on a list of reasons not to use IPv6. If
they also believe IPv6 will be used to any extent, and want to compete
for bits (=dollars) they'll have to work on that reliability.

> > Maybe in six months,
> > conditions will have changed enough for you not to worry.
> 
> Not possible. Not even in 3 years. 

I'd like to dig into that denial some more, preferably without argument
by toilet analogy. :-)
Your objections, as I recall (maybe you could list them again, so we
can discuss what needs to happen on each one):
1.  IPv6 transit is unreliable
2.  An dual-stacked client with IPv4-only connectivity may try IPv6
first, and wait for timeout before successfully using IPv6.

IPv6 transit may not be as reliable as IPv4.  If that's true, I think the
reasons are:
1.  IPv6 has not been properly "productized," so it doesn't have the
same level of monitoring, alerting, and general urgency as IPv4.  This
will change as revenue is associated with it (i.e., as people who bill
per bit notice they're losing bits), and as more traffic uses IPv6.
2.  IPv6 has not been deployed by enough reliable transit providers,
so you don't have as many choices for connectivity.  
3.  Those who do offer IPv6 transit don't have enough peers yet to
ensure connectvity.

Those could all be resolved fairly quickly.  None of them is a three
year project.

> > Question born of ignorance:
> > can you track unique web views if thousands of users are behind a
> > single IPv4 address?  It might not be an issue for your servers, but
> > many websites use that as their primary metric for selling ads, don't
> > they?
> 
> Depends on your methodology. comScore, for example, offers prizes to

So, sometimes yes, sometimes no?
Sounds like that might still be important.  Applications, including counters,
that use IP address will have to use IPv6 address or figure out how to sort 
through NAT.  I would think that content providers who use those 
applications (or the application developers) would be lobbying for 
deployment of IPv6.

> You also have to understand that the metrics folks aren't interested
> in use by machine so much as use by *person*. Users have demographics
> and areas of interest. Machines are mostly just machines. NAT or no
> NAT, tell me how you separate Mom, Dad and Child on the family PC
> evaluating only the packets.

I don't know, that's why I was asking.

 
Lee