Re: Multicast log question
- From: "Ravi Raman" <RaviRaman@xxxxxxxxxxxxxxxxxxxxxxxxx>
- Date: Mon, 1 Aug 2005 23:19:10 -0700
Mike,
I am not quite familiar with the logging, but based on what you are seeing,
I am guessing that c-pkts-lost-cont-net is probably being counted separately
from c-pkts-lost-net though I have no way of confirming that.
Also, the WMS Logging Model document at:
http://www.microsoft.com/windows/windowsmedia/howto/articles/LoggingModel.aspx
says this about c-quality:
• c-quality. The minimum quality experienced by the client throughout the
session; not the average quality. For example, if a client is streaming for
ten minutes and there is a bad network connection for ten seconds, the
c-quality value drops to 60 percent, and it doesn't go up again for the
current session. So the c-quality value might be misleading sometimes. You
should also look at the lost and recovered packets values, the total buffer
time, and the buffer count to make a proper decision about the quality of the
streaming. Don’t just rely on the c-quality value alone.
My guess is that when the client lost 13 packets that weren't recoverable
(i.e., more than 2 packets in a chunk). The quality could've gone to 94 and
would be set at 94 after that. Rest of the packet losses were single packet
losses and all were recovered. Again, just a guess at this point on my part.
Also, I am wondering if a loss of 3 packets at 10 different times, would
count as 30 lost-cont-net packets - in which case the quality will only dip
to 94 and not much lower (which may happen with a straight 30 packet loss,
I'd imagine). Again, just a guess on my part.
Thx,
Ravi
--
This posting is provided "AS IS" with no warranties, and confers no rights.
"Mike Lowery" wrote:
>
> "Mike Lowery" <selfspam@xxxxxxxxxxxxxxxx> wrote in message
> news:ugcKbrfkFHA.3960@xxxxxxxxxxxxxxxxxxxxxxx
> > I've got logging set up on my multicast publishing points and have received an
> > entry that confuses me. The following values are present:
> >
> > c-pkts-received: 10354
> > c-pkts-lost-net: 67
> > c-pkts-lost-cont-net: 13
> > c-pkts-recovered-ECC: 67
> > c-quality: 94
> >
> > Based on past testing we've found that WMP can only recover one lost packet
> out
> > of 11 (the 11th is a Forward Error Correction packet.) The above data
> suggests
> > that all lost packets were recovered, but at one point there were 13 lost
> > consecutively. Additionally, the quality value dropped to 94%. How could
> this
> > be if all packets were supposedly recovered? Are the 13 consecutive lost not
> > part of the 67 total lost?
>
> Nobody has responded so apparently this is as confusing to everyone else as it
> is to me. Here's more confusing stats:
>
> c-pkts-received: 19143
> c-pkts-lost-client: 0
> c-pkts-lost-net: 92
> c-pkts-lost-cont-net: 27
> c-pkts-recovered-ECC: 92
> c-quality: 94
>
> If WMP was able to recover all 92 packets (not sure how this could be with 27
> lost at once), why wasn't my c-quality stat 100%?
>
>
>
.
- Follow-Ups:
- Re: Multicast log question
- From: Mike Lowery
- Re: Multicast log question
- References:
- Re: Multicast log question
- From: Mike Lowery
- Re: Multicast log question
- Prev by Date: RE: Can Windows Media Server 9 accept and broadcast a MPEG-4 video str
- Next by Date: ADS through a firewall
- Previous by thread: Re: Multicast log question
- Next by thread: Re: Multicast log question
- Index(es):
Relevant Pages
|
