[dnssd] DNSSD: DNS-SD discovery for BRSKI and variations

[Toerless Eckert <[e-mail]>]

Dear DNS-SD WG

I was wondering if i could bother you folks in taking a look @ and providing
feedback suggestions 4 our ANIMA-WG draft:

https://www.ietf.org/archive/id/draft-ietf-anima-brski-discovery-05.html

This should have a couple of aspects of general interest to DNS-SD enthusiasts, but
also some new concepts.

If there is interest, i would be happy to present about it at the IETF121 DNS-SD WG meeting.

BRSKI is the IETF ANIMA secure onboarding protocol for devices, where
for resilience and automation its highly beneficial to discover onboarding
servers (registrars), and in the absence of full routing also proxies for them.

If this sounds boring, consider that unfortunately several industry groups have diffeent
opinions about protocol details, so we have ended up in a set of variations of the
protocol where not necessarily all servers are compatible with all clients. So this
draft introduces an extensible method to indicate supported variations so clients
can pick the right server. More importantly, proxies can discover all possible
variations even future ones and create appropriate proxy announcements.

Of course, we want discovery to be fast and resilient, so there is also text about the
details how to select the best server based on prio & weight and time out in case it's
not responding. And how to optimize this in the face of having to do this as a proxy

If that's not annoying enough, then there is also no consensus on what discovery protocol
is the best, so we have to support DNS-SD, GRASP and CORE-LF... today, tomorrow may be
more, and i really don't want to see specs over specs written for a full matrix, so the
draft also attempts to reduce this problem into a cross-discovery mechanism IANA registry,
so that we hopefully can easily define extensions mostly only through additional registrations
in that registry. Which might also be a concept for other protocols with similar interop issues.

If thats' not enough, we also needed to discover client devices (which we call pledges)
via DNS-SD by their serial number, so we had to define a scheme by which we do
that, which is also described.

So, if any of this sounds like an interesting application use of DNS-SD that you'd
like to check out as DNS-SD folks, please do so!

Cheers
    Toerless