Minutes of the Meeting: MMUSIC working group at IETF 83
The MMUSIC working group of the IETF met at IETF #83 in Paris, France. The WG met on March 26, 2012 from 9 to 11.30.
The meeting was chaired by Flemming Andreasen and Miguel A. Garcia.
Paul Kyzivat, Christer Holmberg and the chairs took notes. Hadriel Kaplan acted as a jabber scribe.
This meeting was broadcast live and recorded by the Meetecho team. The recording of the session is available at the following URL:
Introduction and Status Update (Chairs)
The chairs presented the agenda:
No agenda bashing was needed.
The chairs presented the status of the working group (see slides):
Published RFCs since last IETF
- : TCP Candidates with Interactive Connectivity Establishment (ICE)
No other documents are in the RFC Editor, IESG evaluation, publication requested, or WGLC completed states.
Progress of other work items
– RFC 4566bis, draft-ietf-mmusic-rfc4566bis-05
The document is done, we have kept it open for a while, but there are no foreseeable changes.
– RTSP 2.0, draft-ietf-mmusic-rfc2326bis-29
The document was WGLCed for versions -22 (September 2009) and -23 (March 2010). The current version -29 will start WGLC right after the IETF meeting. The focus of this WGLC should be the changes since version -27.
Mostly done, should go to WGLC right after RTSP 2.0.
– SCTP media in SDP, draft-ietf-mmusic-sctp-sdp-01
The document is done, waiting to start a WGLC.
– CS descriptions in SDP, draft-ietf-mmusic-sdp-cs-10
This draft has one issue to solve, related to the format of the correlation token if included in the User-to-User Information element. Paul Kyzivat had a comment on the list suggesting using the same format, since it is the same value, that is used by draft-ietf-cuss-sip-uui-isdn. Keith Drage indicated that there is no requirement for these two formats to be the same. Paul and Gonzalo made argument that we need to justify being different, not being the same.
- Miscellaneous Capability Negotiation, . The draft was just submitted as WG item the same day of the meeting. The document is considered to be done.
– Media Path Middleboxes,
¥ Do we want to keep the latching text in here or refer to the latching draft?
¥ Other than that, the document is done.
Hadriel Kaplan indicates that he has concerns about the goal of this document and the potential target. He sent an e-mail to the list some time ago, and he will be resending it for comments.
– SDP Parallax attribute, and SDP 3D format,
¥ IPR of both drafts disclosed after adoption as WG items
¥ A bit more discussion after the WG status.
– Special mention: Media level ice-options SDP attribute, . The author cannot promise to work further on it. There was a request for level of interest. Cullen expressed interest, but thought should be split into multiple drafts.
The chairs announced that the SDP directorate has been formed, http://www.ietf.org/iesg/directorate/sdp.html. Two reviews have been carried out already.
Discussion on IPR statements related to 3D and parallax drafts
The chairs continued the presentation supported by the same slide deck used for the WG status, http://www.ietf.org/proceedings/83/slides/slides-83-mmusic-2.ppt.
The chairs indicated that both the Parallax attribute in SDP and SDP 3D format, had IPR disclosed by the authors's affiliated companies some time after both drafts were adopted as WG items, and long after the first version of the individual drafts was published. Due to the previously unknown IPR disclosures, the WG needs to re-evaluate the direction that MMUSIC wants to take with respect to the affected drafts. The documents were adopted as WG items at IETF 82, and at that time, no IPR disclosure were known.
Gonzalo Camarillo stated that this appears to be a violation of RFC 3979 section 6.2.1. So he proposes to reevaluate the WG direction on these drafts. Cullen Jennings would like to take time to find alternatives. Keith Drage thinks there should be time to see if anyone else wants to submit an alternative solution. Stephen Wenger noted there is overlap with work done elsewhere (mpeg) and we do not need to do this work at all. Keith agreed that deciding not to do work is a reasonable approach. If the expertise is elsewhere, then those groups can register suitable SDP attributes. Cullen suggested that we un-adopt these as WG drafts, make them back to individual drafts. Roni Even questioned why we would do that
(is it because of the IPR, or because we do not have the expertise and others do?). Stephen argued that the IPR changes the economics; the assertion had been that we could do something cheaper to implement than other standards bodies. Keith says that the best thing we can do now is reverse the adoption, and then later we can reconsider adopting something. The chairs will ask the authors to resubmit as individual submissions again and will remove these as WG items. We will have another decision whether to adopt these at the next meeting. Linyi Tian asked why we cannot keep as WG items while alternatives are considered. But chairs prefer to go this way. The milestone will not be affected for now.
Multiplexing Negotiation Using SDP Port Numbers (Christer Holmberg)
Christer presented his slides.
Roni Even commented on parameters that are affected by bundle. He said that if two video streams are bundled, then the total bandwidth may be different than the sum of the individual streams. But this had something to do with session level parameters, which is not the same. Christer concluded that there is need for further discussion.
Colin Perkins commented on RTCP bandwidth that it is affected. James Polk mentioned that Magnus has draft to specify minimum and maximum, but it might be affected. Harald Alverstrand says we cannot handle all parameters in all drafts, but he wants people to point out any that they are aware of.
Flemming Andreasen commented that this is the biggest issue (e.g. consider sdescriptons.) There must be rules for these. What are the rules for offer/answer. He says that if we try to solve this by identifying individual parameters it will not be good enough. We must come up with categories of parameters that are handled equivalently, so all parameters can be categorized.
Roni says we have rules for session level and media level. This is essentially adding a third level, and we do not have rules for that.
Hadriel Kaplan says we discussed last time that there were issues doing things this way that we would not have if we multiplexed at a lower level on the port, giving multiple RTP sessions.
Cullen requested more time to discuss this draft in the meeting, because it is the most important issue for the group. Chairs said more time have been already allocated, and chairs will not cut the discussion.
Colin commented that there will be restrictions going this way. He thinks the restrictions need to be specified someplace else. Harald disagreed; he thinks this is an SDP issue. We just need to figure out how to signal a single RTP session that contains multiple media types. Colin still says we need a new AVT document about this. Roni says this will be the first topic in the AVTCORE meeting.
Roni says bundle should cover how to multiplex audio and video, etc. , but that there is also a more general need to discuss how to multiplex multiple SSRCs on an RTP session.
Nobody saw reason to mandate rtcp-mux
Signalling Media Stream ID in the SDP (Harald Alvestrand)
Harald presented his slides.
Cullen comments that he thinks there is no need to have separate stream - CNAME has the proper semantics. Harald commented in a way that was not understood by Cullen and others. Magnus gave an example. Cullen says that if two things have the same sync source then they will be placed in the same media stream. Magnus says there are ways to use CNAME, but they will impose limitations on the API. Cullen says those are currently planned, but others disagreed. Cullen says a media stream is a synchronization context, and Harald says it is not. Flemming asked that this be clarified.
Roni asked why the grouping framework is not used. There is a question whether this can work with SSRCs. Harald is concerned that the media stream track value is so specialized it would not fit in grouping framework.
Jonathan Lennox said SSRC grouping is defined as part of the a=ssrc attribute, but it is only within an m-line. Cullen commented that this (A/V)nnn thing was too specific to a particular API. Several other people agreed.
Jonathan asked if this mechanism could work for anything other than
RTCweb. The answer was yes. Harald took away the recommendation to generalize this.
An Extension to the Session Description Protocol (SDP) for Media
Loopback (Hadriel Kaplan)
Hadriel presented his slides.
Hadriel indicates that version -18 was just submitted earlier that morning, and resolved most of the outstanding issues. The only things left are non-technical. The main technical change was to remove media formats from loopback-source and loopback-mirror. Nobody objected to that.
Partha questioned constraint that mirror must be an RTP endpoint.
Hadriel said that happened before he got involved. Magnus said reason is that if you just sent back exactly what you received, then a real receiving RTP stack would find all kinds of errors, e.g., collisions.
Hadriel said he thought doing the RTP this way is what had to be done for the draft to approved, but in reality they are going to implement to just echo without going thru and RTP stack and he suggests later writing a separate draft to explain how to do that.
It was agreed that the draft is ready for WGLC. The draft needs IANA types review, and this will be done in parallel with the WGLC.
The 'trafficclass' Attribute (James Polk)
James presented his slides.
Paul Kyzivat commented that instead of defining the values in the ABNF an IANA registry could be created.
People were requested to review the document.
SDP Media Capabilities Negotiation (Flemming Andreasen)
Flemming presented his slides.
Flemming requested that the new offer/answer procedures be carefully reviewed. Cullen and Christer volunteered. The chairs requested that these be done within the next four weeks. Cullen and Christer committed to that.
Flemming thinks that there are no implications to bundle. But there are issues for bundle that are not specific to this. Jonathan Lennox agreed.
The document is ready for WGLC after the offer/answer review.
Latching: Hosted NAT Traversal (HNT) for Media in Real-Time
Communication (Emil Ivov)
Emil presented his slides.
Cullen said he has commented many times on the security problems due to port scanning. Emil replied that it is discussed in slides and the draft.
It was clarified that the intention is not to recommend the usage of the latching mechanism, but just the publication of this document as an informational document.
Cullen said that restricted latching has problems with carrier grade NATs. Hadriel said that a variation on restricted latching solves that.
Interest was sampled on making this a WG item. Jonathan Lennox did not think MMUSIC was the right place. Magnus suggested the BEHAVE WG. Hadriel said that this was submitted to the MMUSIC WG because the chairs requested it. Cullen expressed thanks to authors for writing the document. There is support for this work to continue. Chairs will work out with the ADs the best way for moving the document forward.
Multipath RTP (MPRTP) (Colin Perkins presenting for Varun Singh)
Colin presented his slides.
As the draft is not addressed to MMUSIC, no decision was made in the WG regarding the future of the draft. There was, though, a general discussion about ICE and finding extra candidates.
Christer commented that the suggested in-band mechanism could cause problems because it bypasses the offer/answer. Colin answered that should couple, so next time an offer/answer is done, the endpoint should include all the addresses that are being used.
Flemming agreed with separating this draft into a core multipath doc, a separate one for ICE, others with either RTP specific or SDP specifics.
Update on Candidate Address Selection for ICE (Ari Keranen)
Ari presented his slides.
It was clarified that the intention of the draft is to deal only with address selection issues; it would not be a general "ICE fix" document.
Jonathan Lennox asked if new candidate list is subset of existing list.
More discussion is expected on list.
RTSP Extension for Substream Control (Peiyu Yue)
Due to lack of time, this draft was just briefly presented (see slides). The discussion was deferred to the mailing list.
Security Descriptions Extension for Media Streams (Sujing Zhou)
Due to lack of time, this draft was just briefly presented (see slides). The discussion was deferred to the mailing list.
The meeting was adjourned at 11.30.