Re: [xrblock] IESG evaluation comments for draft-ietf-xrblock-rtcp-xr-meas-identity-07

[Glen Zorn <[e-mail]>]

On Wed, 2012-07-04 at 10:19 +0800, Qin Wu wrote:

...


> 
> > Second, the cumulative duration is defined in section 4.1 as follows:
> > 
> >   Measurement Duration (Cumulative) : 32 bits
> >      The duration, expressed in units of 1/65536 seconds, of the
> >      reporting interval applicable to Cumulative reports which use this
> >      Measurement Information block.  The value of this field can be
> >      calculated by the receiver of the RTP media stream, for example,
> >      based on received RTP media packets or using RTCP method described
> >      in [RFC3550]. 
> > 
> > The units for the measurement was changed from millisecond to 1/65536 based on working group input. A side effect of this however, is that the maximum cumulative duration that can be expressed is 18.2 hours. This does not seem sufficient, as the cumulative duration may be up to the length of the RTP session.The proposal is to increase the field from 32 bits to 64 bits.
> 
> [Qin]: Agree, we should not limit the cumulative duration to less than 18.2 hours. The proposal looks good to me.
> Here is my input to fix this issue:
> OLD TEXT:
> "
>       Measurement Duration (Cumulative) : 32 bits
>       The duration, expressed in units of 1/65536 seconds, of the
>       reporting interval applicable to Cumulative reports which use this
>       Measurement Information block.  The value of this field can be
>       calculated by the receiver of the RTP media stream, for example,
>       based on received RTP media packets or using RTCP method described
>       in [RFC3550]. 
> 
> "
> NEW TEXT:
> "
>       Measurement Duration (Cumulative) : 64 bits
>       The duration of the reporting interval applicable to Cumulative reports which use this
>       Measurement Information block.  The value of this field is represented using a 64-bit 
>       NTP-format timestamp as defined in [RFC5905], which is 64-bit unsigned fixed-point
>       number with the integer part in the first 32 bits and the fractional part  in the last 32 bits. 
>       It can be calculated by the receiver of the RTP media stream, for example, based on 
>       received RTP media packets or using RTCP method described in [RFC3550]. 
> 
> "
> 

A fine idea, but the question in my mind is why we don't simply
standardize on NTP format for all time values...

...