Re: [dnssd] Web Proxy Discovery on IoT/Matter

[Ted Lemon <[e-mail]>]

Right now the assumption is that if an IoT device needs to talk to a cloud
service, it will be pre-configured with whatever information it needs to do
so. I don't actually know if Matter defines any proxy capability—I think
you'd have to participate in that SDO to find out about that. I work for a
company that participates, but I don't participate directly. Possibly
Stuart knows more.

Thread doesn't do anything like this at all—it's just an IP layer 2 and
layer 3. We specify DNSSD because service discovery is required, and so in
principle a device on Thread could discover a WPAD service on
infrastructure, but I don't really understand how trust would be
established—if you wanted to provide some kind of trust mechanism, you
might be able to leverage DNSSD, but it's not something that's directly
done in DNSSD, nor is there any sort of proxy of the type you are talking
about on the border router as part of Thread. Of course, the border router
is a totally reasonable place to put something like that—it just wouldn't
be done under the Thread rubric.

On Fri, Dec 1, 2023 at 1:34 PM Josh Cohen <[e-mail]> wrote:

> Hi Stuart, Ted, and WG members,
>
>
>
> During the dnssd  meeting at ietf118, the discussion touched on
> IoT/matter, and I mentioned the WPAD discussions and the origins of WPAD
> post sent to intarea.[1][2]
>
>
>
> Is it fair to asset that it is conceivable that IoT devices will be
> connected to networks where a proxy server must be used to access their
> respective cloud services, but must go direct to their local controller
> hubs, or something similar?
>
>
>
> I am curious what, if any, solution is being used by matter to discover
> web proxy servers, or if that is a TBD?
>
>
>
> For discovery, what would be the preferred solutions dnssd/svrloc, dhcp,
> or other on these devices?
>
>
>
> I am putting together an update of the WPAD draft to modernize the
> discovery schemes, and am interested in learning more about the relevant
> scenarios in the IoT space.
>
>
>
> From a config file format POV, running a JavaScript function may be
> infeasible, especially if the device needs resources for whatever the
> functional purpose of the device is.
>
>
>
> So, the JSON PVD format might be easier to swallow than the JavaScript PAC
> file.
>
>
>
> When dealing with Thread, is this scenario somehow handled by the border
> router?
>
>
>
> WPAD origins Email:
> https://mailarchive.ietf.org/arch/msg/int-area/IHzF0V7nPNlfgQjiU7utUo85ngU/
>
> PDF:
> https://mailarchive.ietf.org/arch/msg/int-area/IHzF0V7nPNlfgQjiU7utUo85ngU/2/
>
> ---
> *Josh Co*hen
>
> _______________________________________________
> dnssd mailing list
> [e-mail]
> https://www.ietf.org/mailman/listinfo/dnssd
>