[arin-ppml] IPv4 Depletion as an ARIN policy concern

[Matthew Kaufman]

Roger Marquis wrote:
> Matthew Kaufman wrote:
>> And we most certainly do need a way to free up some additional
>> IPv4 space
>
> Perhaps a rhetorical question but how would you propose getting back the
> legacy /8s, doled out in the dozens back in the day.  Virtually none 
> of the
> owners, organizations like Cisco and HP, really need more than any other
> company their size (i.e., a few /24s).  Then there are all the legacy
> owners of /16s who aren't using a fraction of their addresses.  It is not
> even difficult to identify the squatters thanks to a strong correlation
> with the date of their arin whois record.
Money. When the money available to Cisco and HP for consolidating to a 
few external /24s and renumbering the inside to RFC1918 by selling a few 
/16s is enough that the CIO sees that as a good way for the IT guys to 
be spending their time, that's when it'll happen. Even better is if the 
money is also sufficient to finish the IPv6 upgrades.
>
> Personally I think the outcome is inevitable, as soon as a few 
> politically
> savvy businesses start feeling the budget pinch from IPv4 resale pump and
> dumpers, their lobbyists will be all over the US Congress.  Congressional
> aids will read this lists's archives, and all the plotting over IP resale
> values, and laws will be passed accordingly.
The number of companies with IP space they can part with once it makes 
any financial sense is a lot bigger than the number of companies that 
are growing ISPs in need of more IPv4 space even as they go dual-stack 
with IPv6. There might be lobbying, but it might very well be orthogonal 
to what you're thinking.
>
>> as it has been clear for over a year now that the transition was
>> started too late by almost everyone.
>
> I don't know about that.  The transition was started in time but has been
> stonewalled by those planning to monetize their IP real-estate.  The
> stonewalling has been in the form of continued FUD regarding IPv6 
> NAT.  It
> has also been slowed by short-sighted implementors who fail to see that
> there is no value in IPv6 until a v6 node can access 100% of the IPv4
> Internet as well.
That depends on whether or not the v6 node is dual-stack or not. If the 
latter really mattered, then the IPv6 protocol design would've put 
legacy compatibility a lot higher on the list, right? (ahem)

I don't think it is "stonewalled" so much as "doesn't matter yet" to 
almost everyone's brain. It is very hard to focus on quarterly financial 
results and following SOX-compliant IT proceedures than it is to worry 
about IPv6, especially when the IPv4 Internet is working so well for 
everyone.
>
> The bridge from v4 to v6 has only two real obstacles: 1) a standardized
> version of IPv6 NAT, and 2) a 1:1 mapping of legacy v4 routing to v6.  
> But
> you won't hear much about these two roadblocks in this forum due to the
> signal to noise ratio, skewed by planning (sometimes salivating) over the
> coming v4 resale market.
The former is being talked about in BEHAVE, but did suffer from getting 
shot down in the first round by the belief that the transition would 
take place in such a way that dual-stack would work forever. The latter 
doesn't make any technical sense at all, given how IPv6 was designed.

Matthew Kaufman