[ppml] [Re: IPv6 addresses really are scarce after all]

[bmanning at vacation.karoshi.com]

On Fri, Aug 17, 2007 at 04:01:01PM -0700, David Conrad wrote:
> On Aug 17, 2007, at 2:37 PM, bmanning at vacation.karoshi.com wrote:
> > viable == can actually work in practice.
> 
> Not a good definition.  "Can actually work in practice" for whom?

	good for whom?  :)   perhaps the distribution method should
	be tempered by a caveat that the IP addresses so distributed
	would be used for their "best and highest" use - which in my
	mind is to be used for sourcing/sinking IP datagrams

> Sparse allocation of /48s on demand to anyone willing to pay a yearly  
> registration fee is a perfectly viable distribution model.

	i'm ok w/ that  -  on the presumption that "anyone" will 
	be using the prefix to source/sink datagrams.


> Allocating /12s to national PTT or equivalents in each country on the  
> planet and letting them deal with allocations in country is a  
> perfectly viable distribution model.

	ok... but will temper this w/ the concern that PPT or equivalents
	tend to focus on fiscal gain over productive use.

> Auctioning space to the highest bidder is a perfectly viable  
> distribution model.

	which begs the question of the value basis.  if the IP space is
	treated as commodity - then is it being put to best use?  
	

all these varients are distribution based on available cash - completely
abandoning the historical basis of need.   or perhaps redefining need?

to borrow from an earlier posting in this thread... these are all varients
of the catagory *greed* as a distribution basis.

i've heard three basis for distribution and raft of varients thereof.
perhaps this si the "large number" that you refered to?

> Etc.
> 
> All of these "work in practice".
> 
> What is most likely _not_ going to work in practice is allocating  
> blocks to ISPs and requiring all enterprises obtain PA address space  
> from those ISPs.  Given the liberalization of PI policies either  
> implemented or being discussed in all the RIRs, this would appear to  
> be understood.

	so the IETF architectural model (PA) is flawed in practice and the
	RIR community is coping by discussing/adopting varients (PI).
	The fundamental underpinnings of PI are still need/merit based and
	not on some uncertain "market" valuation. (as described above).

	Do you think that the ARIN community will be happy with any of the
	distribution varients you list above?

--bill
> 
> Regards,
> -drc