[arin-ppml] IPv6 adoption, map-encap for IPv4?

Alexander, Daniel Daniel_Alexander at Cable.Comcast.com
Fri Jun 13 12:09:43 EDT 2008


I hope you make it to the ARIN meeting in LA this fall, and have a
chance to talk directly to many of the people on this list. These are
discussions that should continue. As I've been reading this thread,
however I can't get past two concepts. These are that you are of the
opinion the IPv4 Internet is underutilized because experiments can only
ping ~100 million IP. The other is that IPv6 deployment is a failure
because its not already deployed. 

People continue to think that someone will be able to "see" the whole
(I)nternet. That it is a static, while growing, thing that can be
mapped. Since everyone likes to use Comcast as a reference, here are
some facts. Comcast has more than 38 million active interfaces using RIR
allocated IP addresses. If you factor in re-use of private networks,
another 35 million IP are being used of RFC1918 space. Your thoughts on
utilization would imply that Comcast makes up around 37% of the
Internet. I can't speak with authority, but I'm quite confident that is
not the case. Also keep in mind what you consider "utilized". If you
have a VLAN with a /26, and 50 interfaces connected, are the 50 IP
utilized, or are all 64 now unusable anywhere else on the network. The
38 million number I quote does not include subnet loss, aggregation,
capacity maintenance, or deployment plans. Also consider, we have had
arrangements with other providers, where we routed RFC1918 space outside
our AS boundries. My point to this paragraph is simply the Internet is
much more utilized that anyone can, or will ever see. 

Paul made a good point that applies to IPv6 deployment. It didn't matter
that you had a week to study for the final exam. Other items were
prioritized and the studying happens the night before. It is a fair
assumption that there is at least one or two years of IPv4 space left in
the IANA pool. Would it be remotely responsible for any company to drop
everything to ensure IPv6 is already deployed, when Ipv4 space is still
readily available? Now I'm not saying they wait until the last minute.
I'm simply saying many companies will prioritize accordingly. Where your
assumption is off, is that you think no one is, or wants to work on it.
Any company that wants to wait until the night before, will suffer the
appropriate fate in the marketplace. But like the utilization of the IP
address space, no one will never see what most companies are working on,
or how they have prioritized IPv6 deployment, until after the fact. In
the end, the IPv4 space will become fully allocated, and IPv6
deployments will spread like a brushfire.

Of course this should all be considered just another opinion.


Dan Alexander
Comcast Cable


-----Original Message-----
From: arin-ppml-bounces at arin.net [mailto:arin-ppml-bounces at arin.net] On
Behalf Of Robin Whittle
Sent: Thursday, June 12, 2008 11:01 PM
Cc: Paul Vixie
Subject: [arin-ppml] IPv6 adoption, map-encap for IPv4?

Hi Paul,

In "Re: [arin-ppml] Portable address space vs. IPv6 auto-numbering"
you wrote:

>>> while i am not a member of RRG, if the question is drawn as clearly 
>>> as that, my position would be, forget about IPv4.
>>> the internet will have many more than 2^32 devices connected to it 
>>> simultaneously within our lifetimes, and i think we should preserve 
>>> the option of not using NAT in future generations.
>>> therefore IPv4's growth has a glass ceiling formed by its address 
>>> size, and any effort that's put into growing its routing table has a

>>> fixed return.
>> OK - quite a few people on the RRG agree with you on this.
>> Here is a more nuanced version:  ...
> i think that what quite a few people on RRG might be in agreement with

> is the first qualified statement i made, "...forget about IPv4."  
> however, the qualification is important: "if the question is drawn as 
> clearly as that," which it sounds like it isn't.

The text being debated in the RRG is from co-chair Tony Li and is quite


My understanding of this is that it would make it acceptable for the RRG
to work on the IPv6 problem alone, and not the IPv4 problem.  On PP ML
recently, I think Tony expressed a desire to solve the IPv6 routing
scaling problem by change IPv6 (protocol, stack and I assume
applications) to support GSE, and not work on IPv4 at all.

I suggested a longer and more detailed alternative - solving the
IPv4 problem soon and then taking more time with IPv6, which has more
technical alternatives and less backwards compatibility


Tony's text has received more support than mine, though my text was
supported entirely by Bill Herrin (TRRP map-encap developer) and the APT
(map-encap) team supported fixing both IPv4 and IPv6 with the same
approach.  The remaining map-encap developers - the LISP team - haven't
commented on my text, but my understanding of their earlier comments is
that they want to fix both the IPv4 and IPv6 problems.
   So I understand that the map-encap developers in general want the RRG
to recommend a fix for IPv4.

> furthermore, what i mean by "fixed return" is that it's economically 
> damned: the more effort we put into it the more expensive that effort 
> will turn out to have been.  i don't know if the RRG folks are looking

> at this as an economics proposition but if not, there isn't really any

> necessary relationship between what i said and what they might, under 
> different and/or synthetic circumstances, agree with.

The best place to debate Tony's proposal is the RRG list.

Naturally, the more effort the more cost.  My best guess is that a great
deal of effort will be put into keeping IPv4 going, because
IPv4 space will remain the only way to provide Internet services which
people find acceptable.  IPv6-only services simply don't do what most
end-users want and need.  David Conrad listed some major reasons why
most end-users and ISPs don't want IPv6:


not least because "c) Very few of the applications end users want to run
support IPv6".  While I accept what I am told about major browsers and
email programs working with IPv6 - and some games and perhaps P2P file
sharing programs - I would put it another way:

   A high a proportion of ordinary business and home end-users
   would find some application would not work as expected in an
   IPv6-only service, and/or would find that some site, some service
   etc. was not available to them.  This would cause many or most
   of them (probably the great majority) of them to be too
   dissatisfied to continue the service when they could get an IPv4
   service from a competitor.  Also, for those who remained, the
   ISP would face excessive support costs.

> i'm not just picking nits -- your "more nuanced version" does not 
> represent my view correctly, it is not a restatement of any position 
> i've had nor of one i can agree to.

My "more nuanced version" was an attempt to better state my position,
which I know is very different from yours.  I wasn't expecting you to
agree, but was trying to draw out the differences in our perspectives.

>> ... Many IETF folks have had unrealistic optimism about end-users 
>> wanting or needing IPv6 for over a decade. While I know that IPv4 
>> with NAT etc. falls a long way short of the ideal, I still think many

>> IETF folks are unrealistically optimistic about
>> IPv6 adoption in the next 10 years. I think there are plenty of 
>> coping mechanisms for keeping IPv4 tolerable for most users in that 
>> time frame and probably beyond - and these will be cheaper and better

>> for ISPs than trying to sell an IPv6-only service.
> while on the one hand i completely disagree, the bigger issue is, 
> you're just saying what you think, here, whereas in your two previous 
> articles you shared facts.

Apart from quoted text, everything I write is based on what I think.

I think you are most likely to identify my thoughts as facts when you
agree with them.

What I wrote above is about the future.  We have no facts about the
future.  However, I can reliably inform you that it is a fact that I
believe this about the future!

>> The transition mechanisms are not there. The IPv6 Internet doesn't 
>> connect properly to the IPv4 Internet. People like the
>> IPv4 Internet because everyone is reachable via it.
> these are plain facts and i agree. (i wanted to agree to something 
> here.)

Yes, I prefer to agree if it is all possible.

> as to optimism, jean camp showed some S curves in denver describing 
> adoption of technology, and while a lot of folks didn't understand 
> them, they were well argued and well reasoned, and they give *no* 
> cause for optimism wrt IPv6. so, if provable pessimism about IPv6 were

> an argument in favour of prolonging the lifetime of IPv4, you'd need 
> look no further than elmore/camp/stephens:
> http://weis2008.econinfosec.org/papers/Elmore.pdf

Thanks for pointing me to this.  I have only dipped into the text.
Figure 5 shows two S curves of adoption of IPv6 as a proportion of ASes.
The "optimistic prediction" seems to reach 98% in 2011.  The "best
prediction" S-curve shows only a few percent in 2011, and seems to reach
50% around 2040.  Figure 1 looks bleaker still, but does not take
account of IPv4 address depletion.  The faster and slower rising curves
are the result of increasing or decreasing the "follower coefficient" by
one standard deviation.  Figure 5 is based on the IPv6 adoption rate
after the drop caused by the end of 6bone.

> alain durand's current plan is to native V6 to their customer edge and

> backbone, and use native V6 to carry NAT'd V4, and thus get provide 
> stack to the comcast customer base. apple and microsoft, as well as 
> every IP-capable device designed or manufactured in japan, as well as 
> most f/l/oss software, will all use native V6 if it's present.

OK, but I think this falls a long way short of what end-users want and
need in the next few years.

Do the major PC firewalls and anti-virus programs work fine over
IPv6-only?  What about the various IM programs, AIM, Yahoo Chat etc.
Searching adobe.com for IPv6 turns up some pages which suggest that at
least some of their apps use IPv6.  There's a bunch of shareware and
freeware that people like to use.  I wonder how much of that is ready
for IPv6 only services.  There are really strong reasons for keeping XP
and avoiding whatever else Microsoft wants us to use - so it is a
non-starter to expect customers to switch from XP.  That would need a
fix because I understand its DNS relies on IPv4.

I figure most printers etc. only support IPv4, so the customers are
going to be running IPv4 on their LANs and IPv6 to the outside world.  I
haven't tried this, but I imagine the average user is a long way from
being able to use IPv6-only, without noticing any deficiencies.

What about VoIP boxes and programs.  How many of these are ready for
IPv6 at all and how many can actually do calls and conferencing via some
gateway to IPv4 without any user intervention of inconvenience?

> there's a very real deployment scenario involving islands of this kind

> of dual-stack, followed by address space shortages and/or explosive 
> routing table deaggregation, followed by heightened, agitated capital 
> investment in more dual-stack because it's a way to avoid the worst 
> pain of the shortages/explosions.

I think this optimism about early migration to IPv6-only is based on an
unrealistically bleak notion of how difficult it will be to make much
better use of IPv4 space in the next decade or two and on an
unrealistically rosy view of how many end-users won't be troubled by
apps and various Internet services, not least most web-sites, not
working via an IPv6-only service.

I don't see how you hold this optimistic view when you agree with these
statements as facts:

>> The transition mechanisms are not there. The IPv6 Internet doesn't 
>> connect properly to the IPv4 Internet. People like the
>> IPv4 Internet because everyone is reachable via it.

>> I think a scalable IPv6 could be prepared, which sounds like heaven 
>> to many IETF people (though I still think 128 bits is 64 too long) - 
>> but the main population of end-users wouldn't care, since it is a 
>> different planet with almost none of their friends on it yet.
> i think we're going to see IPv6 routing table bloat earlier than RRG's

> work could possibly complete,

But this is not yet a fact :)

> and that that's the real race here, and that any time spent prolonging

> IPv4's doddering old age with double- and triple-NAT is a dangerous 
> distraction. the internet is not just the web, and we can't go on 
> building new kinds of applications if everything has to be some kind 
> of NAT or map-encap or ALG or proxy. IPv4 is headed for end-of-life. 
> let's move on.

I think our discussion nicely illustrates two widely different views
about IPv6 adoption in the next five years.

If someone got a bunch of ordinary home users this year, and put them on
an IPv6-only service with state of the art NAT-PT or whatever, and then
found that 90% of them or more didn't notice much difference, and didn't
make more than one or two support calls, then I would modify my views.

Comcast and maybe other providers have a lot riding on this IPv6-only
service model.  If it is so feasible, I would have thought they would
have done some actual trials with real users by now.
Alternatively, they might have done a careful study of exactly what a
hundred or so home and SOHO customers do with their PCs and Internet
services over a period of 3 months, and then recreated a representative
subset of this in the lab - to see how it flies with their IPv6-only

There is a discussion on the RRG list about how ordinary end-users
might, and to some small extent are, adopting IPv6 such as via a tunnel
broker service, to give their home services a stable place on the Net,
albeit the IPv6 Internet.

However, many DSL modems support uPnP, and this already provides an
automated and robust method by which applications on PCs can punch holes
in the modem's NAT and so have a reasonably stable publicly accessible
UDP or TCP port.  This means they can accept incoming communications,
run a web server, game server, fully functional P2P client/server etc.
Various dynamic DNS services makes these servers etc. easily accessible
and apparently stable to other end-users.
So it is not as if this IPv6 approach is the only way of getting a
stable public IP address for most users who are on single IP address DSL
services with NAT in the modem.

 - Robin             http://www.firstpr.com.au/ip/ivip/

  - Robin

You are receiving this message because you are subscribed to the ARIN
Public Policy Mailing List (ARIN-PPML at arin.net).
Unsubscribe or manage your mailing list subscription at:
Please contact the ARIN Member Services Help Desk at info at arin.net if
you experience any issues.

More information about the ARIN-PPML mailing list