[ppml] Policy Proposal: Definition of known ISP and changes to

[David Conrad]

On Aug 31, 2007, at 6:34 AM, Paul Vixie wrote:

>> I took a few shots for that but it sound like you're offering the  
>> same
>> thought.
>
> the important observation is that ipv4 its start through a swamp  
> and if we want ipv6 to get a start we might have to allow a swamp  
> also.

I believe we're already on that path.  Shall I dust off my suggestion  
of /48-on-demand?

> remembering
> that the ipv4 swamp used to be a huge portion of the total ipv4  
> table, the
> most recent couple hundred thousand prefixes added have dwarfed the  
> swamp.

What is your definition of "the swamp"?

>> Everybody keeps talking about the huge address space for v6 but  
>> all that
>> space is wasted if nobody uses it.
>
> yes.

I'm beginning to believe that IPv6 as it exists is so flawed and the  
business/economic/political factors are such that in the end, the  
only real hope that the routing system won't all come crashing down  
around our ears is some sort of "jack up" loc/id split solution like  
LISP (et al., see the RAM and RRG mailing list archives if this  
reference is unclear).  As such, you could look at all IPv6  
allocations to date as end point identifiers that will (somehow,  
someday) get mapped into provider aggregatable locators.  It's sort  
of like assuming that we'll be switching in-flight to anti-gravity  
before we run out of jet fuel.

Regards,
-drc