Re: [tram] A note from your (so far) friendly AD ...

+1, let us not try to solve problems we are not chartered to solve


On Tue, Feb 25, 2014 at 9:35 AM, Spencer Dawkins <
spencerdawkins.ietf@gmail.com> wrote:

> Dear TRAMsters,
>
> If I might offer a couple of suggestions to a new working group ...
>
> The TSV area diverts QoS discussions to the TSVWG working group, because
> that's where the QoS DSCP expertise lives.
>
> RTCWeb and TSVWG are having a robust and so-far productive discussion
> about QoS for RTCWeb now (by "robust", I mean "involving chairs and ADs for
> both working groups"). That discussion is more likely to be productive if
> RTCWeb QoS topics don't start popping up on other working group mailing
> lists, like this one.
>
> I'll let your working group chairs actually run the working group, but I
> note as more than an interested observer that the TRAM agenda at
> https://datatracker.ietf.org/meeting/89/agenda/tram/ is really tight, and
> I would really be happier if the working group focuses on the currently
> chartered milestones and demonstrates that you can deliver what you're
> already signed up to deliver, before adding more milestones. The rest of
> the IESG would be happier as well.
>
> Thanks, and see you in London.
>
> Spencer, as your responsible AD
>
> On 02/18/2014 03:33 PM, Simon Perreault wrote:
>
>> Le 2014-02-18 16:12, Oleg Moskalenko a écrit :
>>
>>> How to handle the DS and ECN fields is a part of TURN server RFC (see
>>> the section 12):
>>>
>>> http://tools.ietf.org/search/rfc5766#section-12
>>>
>>> And a good TURN server is supposed to implement that.
>>>
>> Well, the RFC just says that the TURN server should copy the DSCP from
>> one side to the other when doing en/de-capsulation. It doesn't say that
>> the server should actually do QoS based on the DSCP. My point is that
>> there's nothing preventing the server from actually doing it.
>>
>> Anyway, I was expecting responses along the lines of "DiffServ doesn't
>> work on the Internet in general." To which I would have replied: "then
>> couldn't we define a STUN attribute for transporting the DSCP in the
>> payload?" Instead of inventing a new taxonomy (audio, video, slides,
>> etc.), why not reuse DSCP?
>>
>> Simon
>>
>
> _______________________________________________
> tram mailing list
> tram@ietf.org
> https://www.ietf.org/mailman/listinfo/tram
>