[arin-ppml] What is a "host"?

[David Farmer]

On 9/20/10 10:53 CDT, michael.dillon at bt.com wrote:
>> So, we really have to go back to basics if we want to do something
>> sensible with v6 policy. There are two foundations for PI addresses:
>
> Then count interfaces.
> In IPv6, addresses are assigned to interfaces, and networks are composed
> of subnets, which contain a bunch of interfaces.
>
> Nobody is going to walk around and count 1000 interfaces anyway. They are
> going to estimate the number based on some other data such as switch ports
> installed and VMs per server or VMs in the customer database.

Isn't that what mostly happens today regarding these requirements in 
IPv4 today?

>> 2. You have a substantial enough system that renumbering to change
>> your upstream provider is a more unreasonable burden than compelling
>> everybody else to carry the route for your PI addresses.
>>
>> Start from those foundations and work forward.
>
> Given that we are talking about IPv6 renumbering, not IPv4 renumbering,
> subnets still seems like a good measure. In IPv6, the subnets are unchanged
> and you just change the network prefix handed out by SLAAC or DHCPv6.

In theory renumbering in IPv6 is easier than in IPv4, but in practice, 
while it is easier, it is not that significantly easier.  Especially, 
when compared to renumbering IPv4 with DHCP, DNS is the biggest hassle 
in both cases.  This might have been different if we were able to figure 
out A6 records, or something like them.

I would like to see a subnet based definition for the second of the two 
foundations that Bill mentions above.  However, I have not seen anyone 
mention a number, or any rationale for picking a number, that has any 
kind of consensus behind it.  Also, a subnet count is still essentially 
an arbitrary number, yes it is a more meaningful number in IPv6 than 
host count, but I don't believe it will be any less arbitrary.

While I dislike using a host count for IPv6, as I stated in my email on 
Friday, I believe maintaining a host count based on current IPv4 policy 
is necessary for the time being, at least until a reasonable time after 
IPv4 run-out is completed.

If we can come to some kind of consensus on a subnet count I would be 
happy to add that into the policy as a fifth clause in section 6.5.8.1 
Initial Assignment Criteria. It will probably be after the up coming 
meeting, as I doubt we can come to a consensus in the next day or two, 
in order get it into the text prior to the 10 day freeze before the meeting.

-- 
===============================================
David Farmer               Email:farmer at umn.edu
Networking & Telecommunication Services
Office of Information Technology
University of Minnesota	
2218 University Ave SE	    Phone: 612-626-0815
Minneapolis, MN 55414-3029   Cell: 612-812-9952
===============================================