Re: [rtcweb] Comments on draft-jennings-rtcweb-qos (Re: Call for adoption of QoS draft)

On Sep 20, 2012, at 2:17 AM, "Eggert, Lars" <lars@netapp.com> wrote:

> Hi,
> 
> agree that the QCI stuff is probably not useful to discuss.
> 
> The DSCP content is almost fully redundant with what's in RFC4594 (which is in the references but not cited). And RFC4594 has a much better discussion on what DiffServ is and isn't.

So this is trying to work within the bounds of 4594 (and clearly needs to reference that) but I think this narrows things down a bit more that 4594 might or at least tries to clarify things a bit . 

For example 4594 says use "Multimedia Conferencing service class is best suited … video conferencing service"

It also says "Real-Time Interactive is intended for … and video conferencing applications"

It also suggest there are times the right thing for video might be "Multimedia Streaming service class"

As a result we see some major vendors shipping video phone like devices using CSx class while others are using AFx set. 

So this may be 100% redundant with 4594 depending on your point of view - or it might not be redundant -  but we need one clear unambiguous thing about what needs to happen. If it can be reduced to two sentences, I'm all for that. 

> 
> For WLAN, 802.11e is what you want to point to, I guess.
> 
> So this doc - if it's even needed - could be MUCH shorter, by basically just saying "read RFC4594 and map RTCweb flows into the classes defined there like this" and do the equivalent for 802.11e.

It also forms the requirement for the three priority levels at a JS layer and pointed out we don't want the JS setting if a stream is says audio or video.

> 
> Lars_______________________________________________
> rtcweb mailing list
> rtcweb@ietf.org
> https://www.ietf.org/mailman/listinfo/rtcweb