Re: [v6ops] Implementation Status of PREF64

[Philip Homburg <[e-mail]>]

In your letter dated Tue, 12 Oct 2021 18:19:14 +0200 you wrote:
>Right.  My mental model would not us "the same /64 that the wifi is
>attached to" but "a chunk of addresses from a second /64".
>
>I strongly dislike the implications of "any larger site that has a wifi
>network needs to have a /52 provisioned, so thousands of wifi clients
>can have a /64 *each*" on address space management on all layers.

I wonder why this is an issue.

By and large wifi devices have MAC addresses from an 48-bit space. So
we should have enough /64s to give every physical wifi device its own
/64.

Another way of looking at it, the traditional allocation plan is a /48
per site. This gives 64k worth of /64s. So with the exception of
particularly big offices, this is enough for almost everybody.

If we go down to a /56 for home use. You have to quite a collector to
have 256 ethernet devices connected to a home network.

Obviously, sites that are big enough that they have more than 64k 
eithernet devices can get shorter prefixes.

I don't think we should allocate a /64 to every host if the host 
doesn't need one. But it seems that we do have enough space to do so.

As a mental model, provide hosts with unique 64-bit addresses using
DHCP. Then a /48 is equivalent to an IPv4 /16 and a /56 is an IPv4 /24.
Even in a datacenter, it takes quite a bit of space and power to 
exhaust a /48 with 64k devices.