[arin-ppml] Policy Proposal: Open Access To IPv6

[Matthew Kaufman]

Garry Dolley wrote:
> On Sat, May 30, 2009 at 07:51:25PM -0700, Matthew Kaufman wrote:
>   
>> Garry Dolley wrote:
>>     
>>> There are not enough /32 subnets to give away to everyone.  There
>>> are as many /32 subnets in IPv6 as there are in IPv4.  Just in IPv4,
>>> the /32 subnet just held 1 address.  We are running out of IPv4, so
>>> by the same token, if we liberally give away /32's in IPv6, we'll
>>> face the same exhaustion.  
>>>       
>> That doesn't follow. We're talking about /48s and single /32s, not giving 
>> out 256-65535 /32s to the typical early entrant entities as was done with 
>> IPv4.
>>     
>
> What part doesn't follow?  There are 4G IPv4 addresses (/32) and
> there are 4G IPv6 subnets of /32 in size.
>
>   
Almost all of the 4G IPv4 addresses (and it isn't even that big, just 
like there aren't 4G IPv6 subnets) that are assigned are not actually in 
use. We're only "running out" because there's networks all over the 
world that have fewer hosts connected than the subnet has room for. 
*Especially* if you talk about the (relatively large) networks that were 
handed out early on. Show me a "swamp" /24 and I'll show you a network 
that doesn't have 254+ hosts attached and powered on.

I'm not the first to ask you to please not confuse network assignment 
density with host assignment density, either.

Matthew Kaufman