Re: [rtcweb] Pictures of congestion control on the Internet - which is more realistic?

[Harald Alvestrand <[e-mail]>]

On 05/01/2014 08:04 PM, Roman Shpount wrote:
> On Thu, May 1, 2014 at 7:08 AM, Tim Panton <[e-mail]
> <mailto:[e-mail]>> wrote:
>
>     Lets not give up on this.
>
>
> Nobody argues we should give up on this, but using an adaptive rate
> audio codec does not remove the need for QOS.
>  
>
>
>     It is also worth noting that cell phones can do this already,
>     selecting the AMR mode according to the link capacity -  so there
>     is a proof-by-example :-)
>
>
> On cell phone networks voice traffic is treated differently then data
> so, in fact, there is QOS in place on the cell data link.
>
> The whole argument is if application level rate adaptation is enough
> or QOS support is needed. I would say rate adaptation helps, but QOS
> is still needed.

I'd put it the other way round - QoS helps (when it works), but rate
adaptation is still needed.

We might actually agree on what we need to do :-)



-- 
Surveillance is pervasive. Go Dark.