Re: [v6ops] WGLC: draft-ietf-v6ops-unique-ipv6-prefix-per-host-02 - multiple prefixes per device

[Alexandre Petrescu <[e-mail]>]

Le 16/03/2017 à 09:26, Gert Doering a écrit :
> Hi,
>
> On Thu, Mar 16, 2017 at 09:20:53AM +0100, Alexandre Petrescu wrote:
>> Touchscreen and other WiFi, DSL and Cable Boxes of today rarely
>> have only one network interface.
> [..]
>> This draft does neither.  It is called "unique prefix per Host";
>
> All the devices you've listed are not *Hosts*.

Then let me claim that Hosts as we know them (computer with only one
interface) is a too theoretical term - we rarely find it in practice.

In this class of Hosts (computer with only one IP interface) one can
find a few Tablets or Game consoles.  Even then, they have USB
potentially with USBnet and thus IP on it.

Is this the audience of this draft (Tablet/Game console)?  I doubt so.

What happens in practice is that often drafts like this get widely
deployed and only then we realize they dont support more than, well,
Hosts with precisely 1 interface.

We should stay away from creating new such deployments.

There is a BCP requiring more than one address per device.

> This is not about "delegating a prefix to a device that works as an
> intermediate router" but about *Hosts*.  If a host acts as a router,
>  it's not the target audience of this draft.

I am afraid most Hosts out there act as Routers, and most Routers as
Hosts at times.

In both cases /64-unique-per-Computer makes no sense at all and should stop.

But yes, I agree that the concept of distinct prefixes per Computer
(rather than a shared prefix among multiple Computers) should be
pursued.  Just make it multiple.

Alex

>
> Gert Doering -- NetMaster
>