Re: [rtcweb] FW: New Version Notification for draft-reddy-rtcweb-mobile-00.txt

["Tirumaleswar Reddy (tireddy)" <[e-mail]>]

Hi Subha/Martin,

Thanks for the comments. Please see inline [TR].

> -----Original Message-----
> From: Subha Dhesikan (sdhesika)
> Sent: Sunday, January 20, 2013 11:53 AM
> To: Martin Thomson; Tirumaleswar Reddy (tireddy)
> Cc: John Kaippallimalil; [e-mail]; [e-mail]
> Subject: RE: [rtcweb] FW: New Version Notification for draft-reddy-rtcweb-
> mobile-00.txt
> 
> Inline as SD:
> 
> -----Original Message-----
> From: [e-mail] [mailto:[e-mail]] On Behalf Of
> Martin Thomson
> Sent: Wednesday, January 16, 2013 3:29 PM
> To: Tirumaleswar Reddy (tireddy)
> Cc: John Kaippallimalil; [e-mail]; [e-mail]
> Subject: Re: [rtcweb] FW: New Version Notification for draft-reddy-rtcweb-
> mobile-00.txt
> 
> There are many characteristics of mobile networks that make poorly designed
> applications perform poorly.  Honestly, WebRTC isn't special in this regard.
> I'd prefer to see something that is less negatively worded that instead aims
> to improve awareness of the limitations of the network and how a well-designed
> application might avoid the problems that these cause.  See:
> http://developer.android.com/training/efficient-downloads/efficient-network-
> access.html#RadioStateMachine
> 
> For example, I find the assertion that is made here, namely multiplexing will
> degrade the quality of experiences, to be specious.
> The assertion is true, but it presumes that multiplexing is or must be used.
> 
> SD: the question is what assumption is made regarding DSCP values for the
> single UDP flow which has multiplexed RTP flows ?

> - If the assumption is that multiple DSCP is used for the single flow, then
> the document states that it does not work in the 3G network. However, the assumption is premature at this stage as it is not a foregone conclusion. 

This is not a problem of just setting DSCP values.

3GPP uses the concept of bearer. Each bearer is associated with a QCI value. For example if we consider a media session with Audio and video media streams. 3GPP has explicit QCI values for each stream : 

[1] QCI value 1 means packet priority (2), packet delay (100ms), packet loss (10 to the power -2) and used for Voice
[2] QCI value 2 means packet priority (2), packet delay (150ms), packet loss (10 to the power -3) and used for Video

In the above scenario two bearers would be required one for audio and the other for video. Each bearer would have to be associated with packet filters (5-tuple) to identify the audio and video media streams. This needs WebRTC application to be aware of QCI/GBR values associated with each stream. The 3GPP network based on the available resources may or may not permit the bearer resource modification request triggered by the UE. If the network does not permit the bearer resource modification then it does not make any difference by setting DSCP values.

These QCI values are mapped by the network to relevant DSCP values. Hence we have recommended in the draft that Mobile devices in 3GPP network should not prefer RTP Multiplexing.
 
> On the other hand with this assumption, the user experience should not degrade
> because it is no different from different flows with their own DSCPs.
> - If the assumption is that a single DSCP for a multiplexed flow, then there
> is a chance that the experience will degrade but then it should not be a
> problem since the DSCP value in TFT for the bearer should match with the DSCP
> value in the RTP packets sent by the UE.

I did not understand your argument. Are you saying though that WebRTC application would request bearer allocation for audio or video stream only but multiplex audio and video streams on the same 5-tuple ?  If so please explain how would it work in this case ?

--Tiru.

> 
> However, if the mobile elements turn down the multiplexing as is suggested by
> the rtcweb usage draft then either of the above scenarios ceases to exist.
> Regards,
> Subha
> 
> --Martin
> 
> On 14 January 2013 06:35, Tirumaleswar Reddy (tireddy) <[e-mail]>
> wrote:
> > Hi all,'
> >
> >
> >
> > This document describes a set of scenarios in which WebRTC
> > applications have problems in Mobile Networks related to QOS, Mobility
> > (SIPTO). Thanks to Magnus, Markus, Dan and Basavraj for their help and
> > comments. The URL of the new draft:
> >
> >
> >
> > http://tools.ietf.org/html/draft-reddy-rtcweb-mobile-00
> >
> >
> >
> > comments and suggestions are welcome.
> >
> >
> >
> > Best Regards,
> >
> > --authors.
> >
> >
> >
> > -----Original Message-----
> > From: [e-mail] [mailto:[e-mail]]
> > Sent: Monday, January 14, 2013 1:40 PM
> > To: Tirumaleswar Reddy (tireddy)
> > Cc: [e-mail]; Ram Mohan R (rmohanr)
> > Subject: New Version Notification for draft-reddy-rtcweb-mobile-00.txt
> >
> >
> >
> >
> >
> > A new version of I-D, draft-reddy-rtcweb-mobile-00.txt
> >
> > has been successfully submitted by Tirumaleswar Reddy and posted to
> > the
> >
> > IETF repository.
> >
> >
> >
> > Filename:   draft-reddy-rtcweb-mobile
> >
> > Revision:   00
> >
> > Title:            Problems with WebRTC in Mobile Networks
> >
> > Creation date:    2013-01-14
> >
> > WG ID:            Individual Submission
> >
> > Number of pages: 14
> >
> > URL:
> > http://www.ietf.org/internet-drafts/draft-reddy-rtcweb-mobile-00.txt
> >
> > Status:          http://datatracker.ietf.org/doc/draft-reddy-rtcweb-mobile
> >
> > Htmlized:        http://tools.ietf.org/html/draft-reddy-rtcweb-mobile-00
> >
> >
> >
> >
> >
> > Abstract:
> >
> >    This document describes a set of scenarios in which WebRTC
> >
> >    applications have problems in Mobile Networks.
> >
> >
> >
> >
> >
> >
> >
> >
> >
> > The IETF Secretariat
> >
> >
> >
> >
> > _______________________________________________
> > rtcweb mailing list
> > [e-mail]
> > https://www.ietf.org/mailman/listinfo/rtcweb
> >
> _______________________________________________
> rtcweb mailing list
> [e-mail]
> https://www.ietf.org/mailman/listinfo/rtcweb