TRAM - TURN Revised and Modernized

IETF 94, Yokohama
Wednesday, November 4, 2015


Administrativia
===============

09:00 - Chairs' update
        Duration: 10 min

Brandon Williams: Sent detailed review of STUNBIS, needs to be addressed. Not ready for last call.
Justin Uberti: STUNBIS also need a mechanism for smaller MAC.

Working-group drafts
====================

09:10 - TURN Server Auto Discovery
        draft-ietf-tram-turn-server-discovery-05
        Presenter: Prashanth Patil
        Duration: 10 minutes

Andy Hutton: Draft already talks about RTCWEB so should mention RETURN. Informative ref.
Justin: RETURN has normative ref to this. This could have an informative ref to RETURN.
(Consensus for informative ref to RETURN.)

Brandon: Needs more explicit text on validation of (D)TLS server certificates.
Prashanth: Ok we can do that.
(New revision needed.)


09:20 - An Origin Attribute for the STUN Protocol
        draft-ietf-tram-stun-origin-06
        Presenter: Alan Johnston
        Duration: 15 minutes

Brandon: This provides motivation to lie, for example to get through a firewall. Third party auth's kid is more reliable. We might not have use cases for ORIGIN that we're all in agreement about the usefulness of.
Cullen Jennings: For lying to be useful to bypass a firewall, both client and server need to be modified. Third party auth is not practical, too complicated. This is useful by itself and as a building block.
Jonathan Lennox: What is the advantage of this over a Host header?
Cullen: They're exactly the same.
Justin: ORIGIN is not very useful, third party auth is sufficient, no implementer feedback available about usefulness of ORIGIN.
Cullen: WG adopted, went through WGLC, was sent to IESG. Now you're not sure if it is useful?
Brandon: The new limitations are what makes it less useful.
Jonathan: Would a Host header work for you?
Brandon: Not really.

Questions from chairs:
    - Nobody thinks this is harmful.
    - Some people think it is useful.
    - Nobody has any ideas to make it more useful.

Hums:
    - Who thinks this is useful: moderate hum (~5 people)
    - Doesn't cover the use-cases anymore: light hum (~2 people)


Drafts with clear corresponding milestones
==========================================

(none)


Extracurricular work
====================

09:35 - Metadata discovery for third party authorized TURN session
        draft-reddy-tram-token-metadata-01 
        Presenter: Brandon Williams
        Duration: 15 minutes

Chris Wendt: This is orthogonal to OAuth and should remain so.
Justin Uberti: What is then envisioned use case? Is it third-party TURN server providers?

Read the draft: 4 people
Thinks this is useful: 1 person

Varun: Question is not whether this is useful, but whether we want this standardized. People will do this in various ways regardless.


09:50 - Ufrag Permissions for TURN
        draft-williams-tram-ufrag-permission-00 
        Presenter: Brandon Williams
        Duration: 15 minutes

Cullen Jennings: Original goal is to ensure TURN cannot be used to run a general-purpose server. This moves us away from that principle.
Brandon: We agree and we explicitly call it out in the security considerations section. It is a tradeoff.
Jonathan Lennox: Do you expect this to be useful for ICE TCP?
Brandon: That is an open issue.

Two people plus the authors have read the draft. Next steps: discuss on the list.


10:05 - Peer-specific Redirection for TURN
        draft-williams-peer-redirect-01
        Presenter: Brandon Williams
        Duration: 10 minutes

Ari Keranen: (Concerned about changes to ICE-bis that this draft requires.)
Jonathan Lennox: Is using the alternate server mandatory?
Brandon: I would expect it to be advisory.

10:15 - EOF