[arin-ppml] Policy Proposal 2008-6:Emergency TransferPolicyforIPv4 Addresses - Last Call

[michael.dillon at bt.com]

> Why not discuss how to avoid it and how to ensure the 
> transition will be as smooth as possible?  

Because that is largely a technical discussion which is
best held in other, more technical forums such as NANOG
and MAAWG, and others.

> Would it not be 
> better if large ISP's were busy making the transition EARLY?  

No. It is better if the transition happens when most people
are ready, i.e. the vendors have fixed the critical bugs, 
and the operators have successfully completed their internal
tests and trials.

Personally, I would like to see some effort made to coordinate
a whole suite of IPv6 Internet service offerings with the 
Olympics in 2012. China did something like that this year,
and the Chinese are demonstrably much further ahead with
IPv6 than the west is. Of course, we have to coordinate a 
lot more organizations, i.e. cat herding, so that takes more
time, but the 2012 Olympics is at about the right time, and
it focuses global attention in a positive way. Even if IANA
runs out of free IPv4 addresses in 2010 and then ARIN or
RIPE run out in 2011, most ISPs will have enough IPv4 addresses
free to support growth for another year or so. Or, they could
launch IPv6 in some limited way to recover IPv4 addresses to
support growth in other services.

> Do people have something against actually planning the 
> transition? 

I suspect that the best way to plan a transition is to gather
all interested parties in a single forum dedicated to it.
If you know of someone who could host an ipv6-2012 mailing
list, that would be a good start.

> > Also, it is worth noting that it is not ARIN's responsibility
> > to transition the Internet to IPv6. That responsibility lies
> > collectively on the ISPs who make up the Internet, many of
> > whom are not in the ARIN region.
> 
> So, I guess it is ok for them to continue with their policies that 
> discourage, rather than encourage, the transition to IPV6?  
> If it's good 
> for GM, it's good for the country.  Who was GM again?

If you can identify a specific area where ARIN policy actually does
discourage IPv6 transition, then I'm sure that everyone would be
receptive to a policy which fixes that. I'm under the impression that
we dealt with most such issues a year or two ago.

> I've actually made some of those calls, and most of the manufacturers 
> are quite confident they'll be ready when IPV6 rolls out.  
> Unless there 
> is a downturn in the economy or something and they have to lay off a 
> bunch of programmers.  But that is unlikely, so, not to 
> worry.  Be happy.

I know of someone in a very large broadband ISP who made such a call
to their supplier of DSL access boxes. The supplier said something
like, "Since our boxes are based on Linux, we are ready to do a trial
whenever you are." Said ISP has now gathered 100 volunteers from their
technical organization and will begin trialing IPv6 broadband access in
February. The name of the ISP and the supplier are irrelevant, since
close to 100% of DSL and cable access boxes are based on embedding
either
Linux or BSD into the device. Both Linux and BSD have had IPv6 support
for something like 10 years now. That includes stuff like IPv6
firewalling
and IPv6 load balancing. Just because commercial vendors don't have it
in their official product catalog today, does not mean there is a
serious
problem. This is a minor issue.

A much bigger issue is backbone router vendors who fully support IPv6 
in their routers, except that they process switch everything, i.e. use
the slowest path through the device for IPv6.

> What would prevent Comcast, Earthlink, MSN, and AOL (just to 
> pick a few) 
> from doing a large-scale roll-out of IPV6 in 1 month?

Management decision making. I believe that Comcast has already rolled
out IPv6, but making something technically possible, and getting 
management support to sell this to customers as a standard service,
are two very different things. The whole NAT-PT, dns proxy, NAT64 
issue does contribute to this problem because from a management 
viewpoint it looks like IPv6 is not fully-baked yet. However, this
is mostly an engineering and scaling issue that some ISPs could work
through in the absence of standards.

> What would make the answer to the previous question "Nothing."?

You will never get rid of the management decision making problem in
your lifetime. Fact is that most managers nowadays, are people who
have MBA training but do not have experience working at the coalface.
They tend to be risk-averse, always trying to delay things, spread 
risk amongst as many people as possible, and wait until somebody 
else pioneers the path.

--Michael Dillon

P.S. the above are purely my personal opinions.