Re: [tram] Alissa Cooper's Discuss on draft-ietf-tram-stun-origin-05: (with DISCUSS)

[Simon Perreault <[e-mail]>]

Le 2015-05-11 21:14, Stephen Farrell a écrit :
> There's also this issue. If user credentials can collide over >1
> "realm" (where authentication realm means in the HTTP basic/digest
> sense) then those are almost certainly human memorable (why else
> could they collide with non-negligible probability?) and in the
> case of STUN/TURN and, in particular, WebRTC, setting up to use
> human memorable credentials is just basically broken given that
> the humans aren't actually logging in to the STUN/TURN server.

In the case where a single entity controls the authentication databases
for all tenants of a single server, you're right. However, that
condition does not hold in general. That is, we really do need to
properly handle username collisions across realms.

That is a pain point that is currently being handled with a number of
operational hacks. Clever username generation is one of them. It does
work for some use cases, but not in general.

What you're describing is current practice, and we don't like it.

I'd appreciate if we could recognize that there is a problem with
current practice, and actually discuss the merits of the ORIGIN
solution. I understand the privacy concerns, but I don't get two things:

1) Are those concerns identical or different from those applying to,
e.g., the Host HTTP header, or the SNI TLS extension? If they're
different, how are they different?

2) Is there a proposal that could fix the privacy issue? So far I've
seen IESG members pointing out the issue, authors acknowledging the
issue, but no discussion of solutions.

Simon