Re: [Add] When Two Worlds Collide

[Brian Dickson <[e-mail]>]

On Fri, Aug 16, 2019 at 12:16 PM Ted Lemon <[e-mail]> wrote:

> On Aug 16, 2019, at 9:08 AM, Brian Dickson <[e-mail]>
> wrote:
>
> Those basically require that there be a single entry point into resolution
> (single semantically at least, but possibly not physically).
>
>
> It would be helpful if we could acknowledge that in the absence of a
> completely locked down userland on a device with secure boot validation,
> this is completely impossible, and any proposal that is based on the
> assumption that it *is* possible is not worth spending time documenting.
> Or if it is worth spending time documenting, I haven’t heard a good
> justification for why that’s so.
>
>
That is precisely why I am proposing forking these two use cases. The
majority of ongoing discussion should be for the other one, the one that
does not meet those requirements.

There is a consequence to that forking: mathematical logic says, given a
rule that "A implies B", and a statement of fact that "B is false", the
implication is that "A is also false".

What this means is, from an enterprise network security perspective, if any
of those conditions is not satisfied (locked down userland, secure boot, no
access to DNS resolution configuration in userland and thus none of the
current ADD implementations), the device needs to be treated as unmanaged
and insecure, and placed on guest networks only.

The other branch of the fork may allow some form of DoT/DoH, albeit with
restrictions on configuration of resolver at a minimum.

My questions and ideas yesterday were an attempt to discover whether it was
possible to combine the two forks in a single solution.

I am acknowledging that it is not possible.

Brian