ietf
[Top] [All Lists]

RE: IPv4 Outage Planned for IETF 71 Plenary

2007-12-19 12:41:18
Pete Resnick wrote:
On 12/18/07 at 1:32 PM -0500, John C Klensin wrote:

Reporters come to our meetings and attend plenaries.
There are members of the reporter community, or their editors,
who like only those stories that they can sensationalize.   For
them, this little "outage" results in one of two possible
headlines:

    (i) Not even IETF can get IPv6 to work seamlessly.
    (ii) IPv6 is so complicated that only the IETF experts,
    struggling mightily, can get it to work in a drop-in
    environment.

Simply reporting on this thread would lend itself to some interesting
headlines, with minimal sensationalization:

"Proposal that the IETF use IPv6 exclusively for 60 minutes causes
widespread panic"

I really don't have a strong opinion on the proposal itself; I can
live without network connectivity for an hour (or I can cheat and use
my EVDO card for an hour :-) ). But this entire conversation has been
exceedingly informative:

Exactly my thoughts as I read the thread (though I have made a point of
having AAAA in my host file for the things I care about, so I am not
worried). It is really bizarre to watch the reaction to the thought that we
might 'need to be serious about using IPv6'. If we could only get the IESG
to get serious about killing off working groups that are still focused on
IPv4 ... ;)


a) As an Apps guy, this talk does not bode well for how seriously
IPv6 has been taken in getting basic infrastructure issues solved
such that applications can run.

While you are concerned about infrastructure, I am -much- more concerned
about lack of app support. If is fairly easy to fix a handful of root
servers, but it is much more difficult to find apps that are capable of
working in an IPv6-only environment, let alone get them widely distributed. 


b) As a user who runs my own little corner of the network, this
doesn't make me sanguine about being able to get my basic services up
and running under IPv6 anytime soon.

I'm somewhere between depressed and amused.

I have all but given up on any kind of smooth transition. The ongoing
stand-off between the ISP's and Content Providers over who will move first
ensures that this will be an ugly last-minute fiasco. The lack of
wide-spread app support further ensures that there is no way to avoid having
the consumer be very aware of which version of the protocol they are
running, and what each product supports. Here again this thread is very
instructive. If the app community had built agnostic apps years ago, the
agnostic stacks in the current laptops would transparently deal with the
outage (assuming the roots get fixed). Instead there is a panic by people
(scanning the list mostly in the apps community) that should have been
serious in the past, but have procrastinated. 

So for that I applaud the announced outage for disturbing the cozy corner of
ignorance that people have been hiding in, but ... as much as I want to see
IPv6 in wide-spread use, I have to agree with John, Dave, and Eric, this is
not the right experiment. It is not right because it does nothing positive,
other than the threat -maybe- spurring some action. A more realistic
experiment would be to run the entire week with a double-nat for IPv4 (and
nats between the access points to simulate consumer-to-consumer
configurations), where the most public one has absolutely no provision for
punching holes (because realistically an ISP is not going to punch inbound
holes for its customers, or allow them to). Also make sure that there is
only 1 public IPv4 address so the issues of port overloading & exhaustion
are completely exposed. The IPv4-forever crowd will see the failings of
their claim that; n-layers of nat will just work because n=1 does for a
limited subset of applications when the end user has control over hole
punching. 

The resulting headline would be something like: The IETF tried to live in
the upcoming world of multiple layers of IPv4 nat and failed ... Those that
didn't want to suffer that fate used IPv6 enabled apps and moved into a
better future.

I am not opposed to doing an IPv6-only network, but that should be well
planned (~ 1 year in advance), and have the expected outcome of IPv6 as a
success, rather than the known outcome of the upcoming meeting where more
than half would just be cut off without IPv4. It should have happened a
couple of years ago so the IESG would know the state of the world, but
better late than never so tell people now that Fall-08, or Spring-09 will
have a significant portion of the network as IPv6-only, giving them
sufficient time to do real preparation, rather than just panic work-arounds.
Leave IPv4 up only for the external audio feeds, and one AP near the support
desk for dealing with a day-job crisis. With enough time to do real prep,
the result of the experiment will have some meaning. Otherwise it is just
going to create negative headlines.

Tony




_______________________________________________
Ietf mailing list
Ietf(_at_)ietf(_dot_)org
https://www1.ietf.org/mailman/listinfo/ietf