Re: [MMUSIC] Questions/Comments on draft-ietf-mmusic-sdp-bwparam-03.txt

[Magnus Westerlund <[e-mail]>]

Hi Emre,

Comments below.

[e-mail] wrote:
> Hi Magnus,
> 
> 
>>The session level value is bit redundant. It is simply the sum of all 
>>media streams and reflect the total amount of bandwidth needed to 
>>accommodate this session. This can of course be calculated by 
>>a receiver.
>>
> 
> 
> Also, in the <draft-ietf-mmusic-sdp-bwparam-03.txt>, in section 6.2.3, Usage rules section, it is stated that:
> "...The session level value, if present, MUST be the sum of all media level values.".
> 
> I believe this is a true statement for constant bitrate streams. But what about VBR cases? At least for RTP usage, the definition of bandwidth is for max bandwidth consumption over a 1 second sliding window, right? Hence, max bandwidth consumption at session level may be smaller or equal to the sum of these max bandwidths at media levels. And if the client is making resource reservation for QoS purposes, then there may possibly be a redundant amount of bandwidth allocation if the session level bandwidth is simply the absolute sum. Therefore, I believe that the session level TIAS bandwidth value MAY be smaller than the sum of the media level TIAS values. 

Yes, you are correct that if one has several VBR streams they can be 
scheduled so that the sum of all the streams has a lower maximum.

The session level application of the TIAS value has been something I 
have been had different views on, during the development of the this 
draft. For a while I had TIAS only as media level attribute. As I see it 
there are three possible ways to go with this issue.

1. Remove TIAS and maxprate at session level. This will remove the 
redundancy compared to ver 03 definition. It will simplify and prevent 
any possibilities for contradictions between the values.

2. Keep it as defined today as the sum of all media streams values. Is 
redundant, has the possibility to be contradicting between session level 
and media level.

3. Define is at being the actual value when all media streams are sent. 
  Will allow for getting a whole session value that are less then the 
sum of media level values. Will be more complex and will require more 
implementation to consider if both values are present. A given value on 
session level will only be true if all media streams are actually used. 
Not known how often you can actually use this.

Looking at this list I now see that using 2 as current proposed may be 
the least good solution. I would like to get more input if they believe 
in 3.


Best Regards


Magnus Westerlund

Multimedia Technologies, Ericsson Research ERA/TVA/A
----------------------------------------------------------------------
Ericsson AB                | Phone [phone]
Torshamsgatan 23           | Fax   [phone]
S-164 80 Stockholm, Sweden | mailto: [e-mail]

_______________________________________________
mmusic mailing list
[e-mail]
https://www1.ietf.org/mailman/listinfo/mmusic