Re: [xrblock] Transport of I.D-xrblock-rtcp-xr-qoe style quality datain IPFIX

[Hendrik Scholz <[e-mail]>]

Hi,

On 07/17/2012 05:30 AM, Qin Wu wrote:
> I am wondering what the goal of this draft is? Is this draft an part of
> draft-akhter-opsawg-perfmon-ipfix or will be incorporated into draft-akhter-opsawg-perfmon-ipfix?
> Would you like to clarify the status of this draft?

The intention was to not merge the drafts.
Aamer will publish an updated version of his drafts with lots of 
additional elements from my side.
The reason not to merge was that the scope is somewhat different and it 
would (and already is ;)) quite complicated to find the right working 
group for the documents. Adding a narrow scope RTP audio quality draft
doesn't make it easier.

> Just have a quick look at this draft, I think the 1st reference has been outdated
> and should be replaced with I.D-xrblock-rtcp-xr-qoe.

Thanks. Will do!

> Also it seems you give a range for each RTPMOSClass as follows:
>
> rtpMOSClass 1: MoS value < 3.10
> rtpMOSClass 2:  3.10<=MoS value<3.60
> rtpMOSClass 3:  3.60<=MoS value<4.03
> rtpMOSClass 4:  4.03<=MoS value< 4.34
> rtpMOSCalss 5: 4.34<= MoS value<=5
>
> I am wondering what the rationale is for such range allocation. which reference are you based on?

This is based on ITU-T G.107 (table B.1) referring to the perceived user 
experience.

> Lastly why should the rtpMoSClass choose 'number of seconds' as unit?

In order to calculate a MOS value multiple seconds worth of data have
to be present to begin with (TM Forum claims 8-10 seconds).
The data format is flexible in regards to as many seconds worth of
data are transported in a single IPFIX record.
Systems may use fixed time slices (e.g. one record for 30 seconds
worth of stream data) or one record per (variable length) stream/call.
An IPFIX mediator/collector or any data warehouse should be able
to aggregate the data a) from multiple records belonging to a
single streams and b) multiple records belonging to multiple streams.

What unit were you thinking of?

Thanks for your feedback,
  Hendrik

-- 
Hendrik Scholz <[e-mail]>