ietf
[Top] [All Lists]

RE: NATs as firewalls

2007-03-07 07:35:30
--On Wednesday, 07 March, 2007 09:55 +0000 
michael(_dot_)dillon(_at_)bt(_dot_)com
wrote:

...
Also, even though there are only 3 years supply left in IANA,
to date none of the RIRs have changed their allocation policies
to deal with wind-down of IPv4 space or scarcity. Certainly
in some regions, there is the expectation that IPv6 will fill
the gap since it has been proven with consumer and enterprise
customers for the past year or two.

As the wind-down of the IPv4 address space gets more public
coverage, and more larcenous providers charge exorbitant fees
for free addresses, I expect to see a rising public demand 
for IPv6. Note that this implies that we have less than 2
years  to get a solid IPv6 SOHO gateway requirements document
out.

Regardless of what the dates are (I believe that specific
estimates in terms of years are interesting but ultimately silly
because 

        (i) there is every reason to expect a run on remaining
        addresses at some point, whether induced by "public
        coverage", "larcenous providers", ISP or RIR anxieties,
        or something else.  There are no good models for
        predicting timing of the onset of that stampede, nor its
        precise behavior, even if it is clear to many of us that
        it is inevitable in the IPv4 address space end-game if
        current trends and policies continue.
        
        (ii) it is reasonable to expect that by the time such a
        stampede gets serious (or somewhat before that), the
        RIRs and possibly ICANN will try to change policies to
        damp it.  Since no one has make public announcements
        about their contingency plans, or even asserted that
        they have them in place, it is not possible to predict
        how effective any such remedies will be in damping the
        stampede effects or even if the organizations involved
        will actually be able to do anything before it is too
        late.
        
        (iii) it is possible to imagine a number of scenarios
        that could start freeing up some of the space that is
        hidden in some legacy Class As (and even some legacy
        Class Bs).  Some of those would have a positive effect
        on address space availability, others would just
        encourage the stampede effect.  As Tony has pointed out,
        it is unlikely that they would make much difference to
        the actual rate of address exhaustion, but they could
        contribute to, or slightly damp, the tendency toward
        hysteria.  And there is no way to predict which one,
        much less how much.  

As I have also suggested earlier, a different way of figuring
out when we have run out of IPv4 space is not to look at when
the last address block is allocated but at when the perception
or claim of scarcity begins to justify bad behavior (in pricing,
protocol design, etc.).  By that criterion, we ran out several
years ago and can stop having that particular part of the
discussion.

But the bottom line, IMO, is that sooner or later, people will
perceive that it may be hard to get the address block after the
next one they ask for.  We can say that day grows closer with
each allocation and each month on Tony's chart (or Geoff's
chart, or any other plausible chart one likes), but we can't
predict when it will come.   When it does come -- when a lot of
people reach that conclusion -- it is reasonable to predict a
catastrophic change in allocation requests and presumably in
allocations.  Once that occurs, the projections based on models
of past behavior are trash and IPv4 space will be in very bad
trouble... probably in a matter of months and not years.

And that makes the "very little time... if one is going to get
started, the time is now, if not sooner" element of both
Michael's note and my recent one the important news for the
IETF.  Or we can leave it to the marketplace (including the
larcenous providers and even more larcenous address-hoarders) to
sort things out for us.

     john



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

<Prev in Thread] Current Thread [Next in Thread>