[arin-ppml] IANA IPv4 /8 burn rate....

Stephen Sprunk stephen at sprunk.org
Fri Aug 29 17:22:51 EDT 2008

Ted Mittelstaedt wrote:
>> -----Original Message-----
>> From: Stephen Sprunk [mailto:stephen at sprunk.org] 
>> Sent: Thursday, August 28, 2008 8:42 PM
>> To: Ted Mittelstaedt
>> Cc: 'Scott Leibrand'; 'Alain Durand'; 'ARIN PPML'
>> Subject: Re: [arin-ppml] IANA IPv4 /8 burn rate....
>> Ted Mittelstaedt wrote:
>>>   Everyone who assumes that moving to IPv6 would be better has I think already provided a boatload of arguments as to why their way would be better.  But I have not really heard any arguments from the people who want to stay with IPv4 as to why their way would be better.
>> I think that answer is simple: the short-term cost of adding more NAT is 
>> lower than the short-term cost of moving everything to IPv6.  There's a 
>> lot of stuff that _still_ doesn't work (well or at all) with IPv6, despite over a decade of work and sweeping claims by IPv6 supporters, so the cost of the latter option isn't even calculable because it's not possible -- but even the parts that are possible will undoubtedly cost 
>> more, in the short term, than just tossing a few more NAT boxes into the 
>> network.
>> I think everyone is in agreement that the long-term costs of IPv6 are cheaper than IPv4+NAT; what we're really debating is if and when that transition will happen and what to do in the meantime.
> If the long term costs of IPv6 are cheaper that is a huge argument
> against tossing a few more NAT boxes into the network.  In short
> you have just successfully argued one of the many points AGAINST
> a liberalized transfer policy, and FOR moving to IPv6 asap, that
> is, why IPv6 is better.

No, I have pointed out the frame of debate; I am not saying either side 
is correct, because the economic and technical factors will vary for 
each organization as well as over time.  That is why I'm leaning 
_towards_ the transfer policy, because it gives each org the flexibility 
of deciding for itself when to make the transition and how, depending on 
how it assesses the factors in its particular situation.

> I had asked for arguments from the people who want to stay with IPv4 as to why their way would be better, and the best you can come up with so far is to take an argument saying the IPv4 way would be worse, and turn it upside down and paint it a different color and hope I wouldn't notice this?

Your question presupposes that there exist people who think it is best 
to stay on IPv4+NAT forever.  I have seen no evidence that such exist, 
therefore I assumed your question was about people who are trying to 
stretch out or delay the transition.

> Surely you do better than that!
> The transition would happen tomorrow if people just went and did the work.

What people?  Which work?

> Unfortunately the IPv6 transition is something that everyone doing it is dependent on everyone else doing their bit.  The end users can't switch unless they get native IPv6 from their ISPs, and they can't use a proxy because an IPv4->IPv6 proxy standard is still under debate.  The ISP's can't switch until their feeds switch, and those can't switch until their peers switch, and their peers are probably the worst of all.
> You get 3 backbones like Sprint, ATT & MCI in the room and MCI will say they can't go to IPv6 until Sprint does, and Sprint will say they can't go to IPv6 until ATT does, and ATT says they can't go to IPv6 until MCI does.
> If you're a father of children surely you will have recognized this as classic textbook BSing by now.

That's not what I see.  The vendors aren't making the products necessary 
because customers aren't asking for them and there is financial 
incentive to _not_ do work that people won't pay extra for.  The 
customers aren't asking yet because they don't want to pay extra for 
features they don't yet need.  There is no need yet because the Internet 
hasn't collapsed yet.

Believe me, I've tried to get my employer to implement IPv6, and the 
reality is that, so far, we have lost zero sales because we didn't have 
it, while other features competing for the same developers _would_ have 
lost us sales if we didn't deliver them.  Until a non-trivial number of 
our customers start refusing to buy products without IPv6 support, it's 
financially irresponsible for us to do the work.  The engineer in me 
doesn't like that answer, but the stockholder in me takes the results to 
the bank...

Calling reality "BS" does not change the fact it's still reality.

> Claims that IPv6 is not ready yet are EXACTLY LIKE claims that Microsoft Windows Vista isn't ready yet.  They are simply bogus nonsense excuses that people make because IPv4 is a comfortable pair of old broken-in shoes, and IPv6 is the brand new pair of shiny, creaky, squeaky shoes.  Yes the new shoes will take some breaking in and you will get some sores for a bit until you adjust.  But how long are you going to keep putting tape or whatever on the old shoes?  Until they fall apart and the Internet stops working?

The economics strongly favor that scenario, and that scenario has been 
played out several times in the past: nobody is willing to step up and 
spend the money to fix things, which gives their non-acting competitors 
an financial advantage, so any major change to the Internet must result 
from actual failure because that forces _all_ players to act at the same 

The financial markets also encourages executives to cut costs today 
despite disastrous long-term consequences, because the executives will 
rake in tens of millions of dollars in the meantime and then pull the 
ripcord on their golden parachute when their choices catch up to them, 
leaving the mess to the next round of execs -- or a bankruptcy court -- 
to sort things out and start the cycle over again.


More information about the ARIN-PPML mailing list