Re: [rtcweb] Comments on draft-ietf-rtcweb-use-cases-and-requirements-07

[Göran Eriksson AP <[e-mail]>]

 

> -----Original Message-----
> From: Cullen Jennings [mailto:[e-mail]] 
> Sent: den 8 juni 2012 14:23
> To: Göran Eriksson AP
> Cc: [e-mail]
> Subject: Re: [rtcweb] Comments on 
> draft-ietf-rtcweb-use-cases-and-requirements-07
> 
> 
> On May 30, 2012, at 5:06 , Göran Eriksson AP wrote:
> 
> >> 
> >> -----Original Message-----
> >> From: [e-mail]
> >> [mailto:[e-mail]] On Behalf Of Cullen Jennings
> >> Sent: den 16 maj 2012 03:31
> >> To: [e-mail]
> >> Subject: [rtcweb] Comments on
> >> draft-ietf-rtcweb-use-cases-and-requirements-07
> >> 
> >> 
> >> Like to add case where application has one two video 
> streams but one 
> >> is far more important than the other. Should be a way to make sure 
> >> that preferential treatment is given the the important stream over 
> >> the less important streams.
> > 
> > I agree that this is a relevant area to address to secure 
> that WebRTC 
> > is competitive compared to apps done in native OS's, 
> especiallin in enterprise context.
> > 
> > The current use case document states something like "being 
> able to use priority functions in network nodes".
> > This is vague since it touches many matters such as whether to put 
> > audio and video on the same IP-flow or have them in different ones, 
> > which is essential when considering mapping on LTE radio 
> channels; the potential use of DS to identify flows, which 
> may be relevant in enterprise context, etc, how to cater for 
> treament of stuff on the datagram channel, multihoming 
> consequences, etc.
> > 
> > I am not sure however how far into such matters we should 
> go in this 
> > WG, but given the importance of making a good solution 
> working across 
> > OS's and browser vendors and access technologies, I am 
> leaning towards support a discussions about such details in 
> this WG for instance using the use case document.
> > 
> > What is Your take of this? 
> 
> I think we need to discuss it in the spec. 
>Let's imagine the 
> case where we want the brows to set one DSCP for audio and a 
> different one for video. I don't think we can have setting 
> the DSCP 100% under control of the JS App as that has 
> security problems. But the browser should set the DSCP to 
> something and if there are multiple video streams, I think it 
> is important to be able to indicate some video streams are 
> less important than others - only the JS app knows which one 
> is less important. 

Agree that the JS controlling the DSCP marking is an issue- I am more leaning towards
looking to solutions putting the "origin" in control. Also, the server side of the "app" should also be able to in a position to understand the relative importance on the different
streams/flows in an application. But this is of course up for discussion.

Another set of use cases could be considering the data on the data channel- we have scenarios where these have a significant impact on the UX and it could therefore be
relevant to consider these flows as well- after all, it may be nice talking to/looking at someone but if we also expect some shared data, I am pretty sure we will spend time complaining on the "darn" network more than doing whatever was the original purpose, :-)!

Going forward, I think Your proposal for adding use cases detailing requirements in this
area is a very good one and a good, next step. I expect this will be brought up in next week interim meeting for a discussion about how to proceed.


> 
> 
> 
>