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

William Herrin bill at herrin.us
Mon Sep 20 11:38:49 EDT 2010


On Mon, Sep 20, 2010 at 5:02 AM,  <michael.dillon at bt.com> wrote:
> It is still not clear whether virtual interfaces on virtual machines
> such as XEN, VMWare, HyperV use a link-local address. Do they count or
> not? It is also not clear whether or not this might be an implementation
> detail, i.e. some implementations of virtualisation might use a link loca
> per virtual interface and others might not. For that matter, it may even
> be a configuration detail.
>
> But let's go back to square one. A single /64 subnet has the capacity to
> hold all the hosts that you or I might ever need, even if we expand to
> become a multinational business empire and implement virtual machines on
> our virtual machines.
>
> So why are we counting hosts? Shouldn't we just be counting subnets as the
> most finely grained measure that an RIR is interested in?

Michael,

If we're going to tally anything, counting the IPv6 subnets is a much
better match to the technology than counting the IPv6 hosts.

But, as many problems as that solves, it creates a whole "host" of new
ones.  Infrastructure providers tend to need a lot of subnets while
hosting providers tend not to. If we went by straight subnet counts,
we'd likely disadvantage the hosting providers.

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:

1. You're multihomed with BGP and that requires BGPable address space
for which there is no evident advantage (and clear disadvantages) to
using a cutout from someone else's addresses.

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.


Regards,
Bill Herrin


-- 
William D. Herrin ................ herrin at dirtside.com  bill at herrin.us
3005 Crane Dr. ...................... Web: <http://bill.herrin.us/>
Falls Church, VA 22042-3004



More information about the ARIN-PPML mailing list