[ppml] IPv4 wind-down

[Jason Schiller]

On Wed, 21 Mar 2007, Martin Hannigan wrote:

> An above ground v4 trade would be helpful. Allowing V4 to be
> treated as property, at least for legacy space, would be
> required. The economics of that are interesting, and the
> outcome predictable. Seems like a smoother transition
> than tossing a bomb into the problem like the aforementioned
> policy seems
> to do, IMHO.
>
> -M<

On Thu, 22 Mar 2007, Geoff Huston wrote:

> "interesting" in that "all markets are interesting" or "intersting in
> that "this market would be unique in a number of ways, only some of
> which are predictable"?

> Again its not clear to me what you mean by 'predictable."  Yes, in
> effect a market for a good prices itself against the cost of
> alternatives, and a market in IPv4 may well perceivce NATs and IPv6
> as alternatives and prices within such an IPv4 market may well
> reflect the perceived cost (and benefit) of such alternatives. My
> question to you is whether this is what you mean by a predictable
> outcome, or are you hinting at some other view of the behaviour of
> such a market?

I have a couple of specific questions about this above ground v4
trade.  

Lets assume for the moment that the property thing is worked out and also
asusme that you can definitively state that a given person has the
authority to sell/rent/what ever a partacular block of IP addresses.  Lets
also assume their is a fair market for IPv4 adresses.

1. Could that person sell/rent/what ever a portion of that network.  Say I
am offically representing IBM, could I sell 9.9.9.0/24 but keep the rest
of the /8 to use for myself... or maybe sell off other /24s at a later
date or to a different org?  Or would I be required to sell the entire /8?

2. If that person put 9.9.9.0/24 up for auction, could they sell it to the
highest bidder?  Or would the highest bidder also need to show justified
need of that space?  In other words does money become the only measure of
if the space is justified?

On Thu, 22 Mar 2007, Kevin Kargel wrote:

> I see a lot of people discussing what each person thinks the rest of the
> world should do.  How about some discussion about what we each ARE doing
> so we can provide some ideas and pro-active community.  I know that I
> could use some knowledge and help tuning my strategy.  
[snip]
> I have made sure that all hardware is IPv6 capable as I replace it.  I
> am doing this in a attritious mode and not expending special budget in
> the name of IPv6.  As most business models amortize network hardware on
> a 2-4 year schedule it is still early enough to plan IPv6 in to a
> networks future without undue budget strain.  If I waited two or three
> years to start this hardware transition I am afraid I might find myself
> in a crisis situation which would be much more costly.

In 2003 we built a separate IPv6 network (AS12702) in Europe.  In 2004 we
built a seperate IPv6 network (AS284) in North America.  To keep costs
down both were deployed as a GRE overlay network over the existing IPv4
network (AS702 in Europe and AS701 in NA).  IPv6 nodes were co-located
with the IPv4 nodes that transported the GRE back-bone links.  Customers
accessed the network through GRE.

After much experience and certification we have recently field tested
dual stack IPv4/IPv6 on AS701 in select locations in a 6PE
configuration.  In the next few weeks we will begin a migration away from
AS284.

After more certification we expect to migrate to dual-stack IPv4/IPv6 in
the core... possibly next year. 

Normal equipment refresh, and other projects driving edge upgrades will
drive growth of IPv6 locations.  The problem is that IPv6 is not
generating new revenue so you can not justify new investments to support
IPv6.  

As a large ISP there are a couple of factors that affect an IPv6 roll
out.  First is the requirement that IP forwarding be done in hardware. 
Software switching 6Mpps is problematic.  Thus upgrading the network to
support IPv6 is not as simple as a software upgrade on some cisco
7200s.  Often we are talking about replacing hardware.

Secondly, due to the size of the network, the number of devices, the
amount of coordnation, amount of remote hands available, maintence
policies, and size of our maintence team, it could take 2-3 years to
upgade every node in the network.  We like to amortize network hardware on
a 5 year cycle.  So my time line is a bit different than yours. 

As for the "crisis situation" at the moment it doesn't look like there is
a significatnt amount of new functionality in IPv6 for people to demand
it.  Likely IPv4 exhaustion will drive IPv6 adoption.  Currently people
are concerned about IPv4 exhaustion, and are looking for an upstream ISP
that is IPv6 capable, has IPv6 experience, and will be there to hold their
hand when they migrate.  Some people are begining to experiment
with IPv6 to make sure their eqipment is capable, to gain experence, and
develop their migration plan.  Beyond that there are a few IPv6
application developers, and gov't site that require IPv6 transport.

So the big question is when will IPv4 exhaustion occur.  Geoff's
projection is 06/2012, Tony's projection is the end of 2009.  Both assume
no run on the bank.

If you expect all new networks after the IPv4 exhaustion date to be
IPv6-only, at some point existing IPv4-only sites may want to talk to
some of these IPv6-only sites (rather than lose customers to their
competitors) and will be forced into dual stack (or requiring some sort of
IPv4/IPv6 proxy).  Lets call this IPv6 critical mass.  

It is hoped at the time of IPv6 critical mass, that all hardware upgrades
will be complete, and all elements will be running dual stack.  Since this
is likely at least 5 years away, and I am at the begining of a refesh
cycle this should not be an issue.

On the other hand is the concern about the routing table size.  If one
projects out the size of the current IPv4 Internet table, adds to the the
projected size of you internal routes, and derive the expeced number of
IPv6 Internet routes (if everyone does dual stack and advertises the same
number of more specifics for TE, but discount the extra routes from the
assigments of dis-aggregates) and project the number of internal IPv6
routes you end up with a scary number... My projection for 03/2014 if
wide spread IPv6 adoption occurs by then is 1.3M IPv4 + IPv6 routes.  That
is larger than the current FIB limitations of routers I can buy today, so
likely my refresh cycle will be artificially shortened.   

 
Sorry I am a little late to the thread, but I wanted to read the whole
thread before posting.


__Jason
==========================================================================
Jason Schiller                                               (703)886.6648
Senior Internet Network Engineer                         fax:(703)886.0512
Public IP Global Network Engineering                       schiller at uu.net
UUNET / Verizon                         jason.schiller at verizonbusiness.com

The good news about having an email address that is twice as long is that
it increases traffic on the Internet.