[arin-discuss] ipv6 technology supplier phone bank?

Tony Hain alh-ietf at tndh.net
Tue Sep 29 13:08:34 EDT 2009

Lee Howard wrote:
> I'm not sure this is still an arin-discuss thread.  I'm happy to move
> it
> somewhere else, unless we quickly get back to an ARIN-member
> question, like, "How should ARIN spend its resources to help?"
> > > ISPs will break IPv4 for their paying customers, and charge to put
> it
> > > back
> > > together?    Brilliant!
> > > Especially if the guy down the street hasn't broken it yet.[1]
> >
> > It is not my plan, it is what the ISPs have backed themselves into by
> > stalling their IPv6 deployments.
> IPv6 in the core is easy(ish).
> IPv6 in the data center is easy(ish).
> IPv6 at the edge is hard.  It's partly a network value problem:
> consumers
> won't pay more for gateways and gadgets that support IPv6, so vendors
> don't make them.

They should not have to pay more, as the actual differences at that level of
device are down in the noise. What they need is motivation, and plumbing is
not motivating in its own right, particularly when nobody is explaining the
problem in terms they can understand. The next 'YouTube' would be one way to
motivate them, if it only worked over IPv6. Of course it would never gain
enough audience to be a motivator without at least a minimal IPv6
infrastructure, so what it will take is something like a new service from
google or Netflix that says 'buy the (App name) enabled home gateway to
enjoy this service'. No plumbing details required. 

> > > If IPv6 requires a new home gateway but IPv4 only requires part of
> a
> > > CGN,
> > > which is cheaper?  Who incurs the cost? [2]
> >
> > If you look at the CGN deployment plans, they all require a new home
> gateway
> > anyway.
> Really?

At this point they basically all boil down to ds-lite, and/or some form of
A+P routing to deal with lawful intercept. Simple cascaded nat is the
solution from people that haven't given serious thought to the design.

> > CGN will break just about everything but simple html and pop, and
> can't
> > begin to scale in the face of the AJAX fad. It really doesn't matter
> which
> > one you pick, you are just trading off which apps break, and how much
> the
> > new home router is going to cost.
> Really?  I'm looking for a list of application protocols somewhere.  Do
> you
> have one handy?

Anything that uses upnp will fail. 

Google-maps will fail if you oversubscribe the addresses by > ~100:1. The
iTunes store will fail if you oversubscribe the addresses by > ~10:1. Future
AJAX based app given that pattern in release dates and number of objects
being pushed at the consumer ??? 

> > > [1]  Notice whether IPv4 is broken for existing customers or only
> new
> > > customers.
> >
> > This is a very short term viewpoint.
> Uh, yeah.  Remember the problem under discussion was that CIOs are
> too short-sighted to deploy IPv6 this early, because they're focused on
> quarterly earnings?

No, cap & grow is an engineering viewpoint looking to do the least amount of
work. The business side of the house will force the issue over time so that
even the older customers will feel the pain, because their goal is to retain
the largest number of customers. The CIO is in the middle, but at the end of
the day closer to the business side than the engineering side.

> > It doesn't take a Nobel prize winner to
> > figure out that the existing customers will be screwed over just like
> the
> > new ones in short order  'just because you can'. There will be lots
> of
> > make-work shuffling the old customers into the new model and back
> when they
> > complain, just to figure out which ones are using apps that care.
> You have a very low regard for ISPs.  My mileage is different.

Just observed behavior, unclouded by the ideals they claim to aspire to.
This is just fundamental business practice, 'retain the largest number of
customers by providing everyone the least you can, and no more than any of
them actually claims to need'. When potential new customers demand public
IPv4 after the reserves are gone, they will be reclaimed from old customers.
The old customers that don't complain will be left in the broken state,
while those that do will be put back. 

> > > What happens to competition if users can't switch providers
> > > and receive a comparable IPv4 service?
> >
> > This question makes it sound like somebody cares about
> competition.... ;)
> > There is not going to be a 'comparable IPv4 service', no matter how
> you want
> > to define that, once the free pool is gone.
> Working from your previous paragraph, there could well be two classes
> of Internet subscribers: screwed and unscrewed.  Could be based on who
> got there first, or could be based on who pays what.

Cap & grow is the engineering approach, poach from the old is the business
approach. From my observations, the business side of the house beats down
the engineering side of the house every time.

> If it lasts for any length of time, we may see broad adoption of
> RFC3093
> (Firewall Enhancement Protocol, i.e., everything over HTTP).

Absolutely, but the wild card here is that the OS's can just tunnel over the
crap at lower layers if the ISPs don't get their act together. 

> > > [2]  Important question.  Also important is, "Who are they paying,
> and
> > > what
> > > does their ROI look like?"  Are CPE vendors thinking, "If I don't
> > > support
> > > IPv6 until 2012, then all my consumers will have to buy new gear
> from
> > > me
> > > again when they're forced into IPv6"?
> >
> > I have no idea what CPE vendors are thinking (including the Linksys
> team
> > which I try to motivate), but I really doubt they will hear IPv6 from
> > consumers, while the ISPs have not been telling the CPE vendors that
> they
> > need IPv6 capable cpe until very recently, and most ISPs haven't done
> that
> > yet. CPE vendors run on very thin margins, so they are not even going
> to
> > start thinking about IPv6 until they get very clear indications from
> the
> > ISPs as to what is needed. So far the few messages I am aware of are;
> not
> > coherent; nothing more than an acronym; have not clear timeframes or
> orders
> > ... YMMV
> My point was more that CPE vendors don't sell to ISPs, they sell to
> consumers, who really don't want to be bothered.  

This is not globally true, but more to the point, the CPE vendors are not
getting a clear indication of what the ISP service interface will look like,
so they couldn't build a box no matter who the final customer would end up

> ISPs can tell what is
> needed (work in progress) but don't generally send POs.

ISP service is regional, while the CPE boxes are generally global. In large
part the ISPs stopped participating in the standards development process,
particularly those in the ARIN region for the topic of IPv6, and now that
lack of effort is coming back in the form of a gap in the service interface

> How should ARIN spend its resources to help?

That is less clear at this point. It would have helped if ARIN had been more
informative about the burn rate on the IPv4 pool about 5 years ago, but too
many people were still in denial to allow that to happen. Maybe providing a
vendor-neutral forum for the ISPs to tell the content providers that 'the
plumbing is moving, so optimal access to the consumer means dual-stacking
the content', would be a place to start. Getting Disney, CNN, Facebook ...
in a room to hear why this is happening would go a long way toward
mitigating the problem here. As you said, changing the data center is easy,
the hard part of that is providing the motivation to do it. 


More information about the ARIN-discuss mailing list