[Megaco] H.248.1 V3 Week 5 - RE: Statistics:nt/os, nt/or;
For information only:
I've mentioned in our "Week 5" comments that ALCATEL will submitt a contribution to next SG16 on "H.248.1 Annex E.11 Network Package - Recommendations for Statistics Usage".
An initial proposal (file Alcatelx1 T05-SG16-050726-D H.248.1v3 nt Package_Ed01.doc , Alcatelx1 T05-SG16-050726-D H.248.1v3 nt Package_Ed01.pdf) is ready and uploaded in /av-arch/avc-site/Incoming.
Prof. Okuba, could you please move the files in directory
http://ftp3.itu.int/av-arch/avc-site/2005-2008/0507_Gen/Working_Docs/
Thoughts, comments?
Regards Albrecht
PS Walter, I did cover your mentioned relation in clause "E.11.5.3.2.2.1". But I didn't consider the correlation of statistics from different Terminations of the same Context so far in the proposal. Perhaps I'll add a subsection on that item ("which is questionable in general for statistics without any time information, like start/end time, or time stamps per probe etc; ... and I'm not aware of the existence of such kind of H.248 statistics so far ..")
"Binz, Walter"
<Walter.Binz@key To: Albrecht SCHWARZ/DE/ALCATEL@ALCATEL, megaco@ietf.org mile.com> cc:
Sent by: Subject: RE: Statistics:nt/os, nt/or; Re: [Megaco] Call Flow Example in megaco-bounces@i H.248.1Corrigendum1 (03/2004) etf.org
23.05.2005 15:37
Albrecht et al.,
my understanding is that nt/or & nt/os for a 64kbps physical termination (such as in the H.248.1 call flow example) should be about 8 times nt/dur (8 PCM samples per ms), independent of the chosen RTP/AVP (I assume that codec and silence suppression functions are part of the ephemeral termination).
Therefore in the H.248.1 Call Flow Example below, nt/or for the ephemeral termination should be about 10 times lower than nt/os for the physical termination (and NOT the same), as a G.723.1 codec was used for the call.
I would appreciate if some guidance text can be added to the nt package.
Regards, Walter
-----Original Message----- From: megaco-bounces@ietf.org [mailto:megaco-bounces@ietf.org] On Behalf
Of
Albrecht.Schwarz@alcatel.de Sent: Friday, May 20, 2005 9:12 AM To: megaco@ietf.org Subject: Statistics:nt/os, nt/or; Re: [Megaco] Call Flow Example in H.248.1Corrigendum1 (03/2004)
Hello Alberto et al.,
I think the question was not the *unit* of the time-based statistic
nt/dur
versus the unit-less statistics nt/or & nt/os. Looking from a high-level perspective on the traffic volume related statistics nt/or & nt/os on circuit- and packet-side on the H.248 Context (MG) in this example, it looks like that the conservation law is invalid ("traffic volume increase from C2P direction, whereas the volume is constant in the opposite direction P2C":-) Of course, whether the conservation law is applicable at all depends on codec type, silence suppression mode, RTP FEC mode, and other configurations.
The crucial point is rather, which protocol layers have to be considered for these volume-based statistics? This basic question becomes more open due to the "generic scope" of these statistics: they are defined for ALL type of H.248 Terminations (physical T.: ALN, TDM; ephemeral T.: RTP,
UDP,
IP, AAL2, AAL1, ATM, etc etc).
This issue was frequently discussed in the past, also in last SG16
meeting.
Just the note therefore that I'm currently working on a contribution on that subject for next SG16. The intent aren't any technical changes to the nt Package, I rather do
see
a need to provide some guidance on statistics usage. Such recommendations may be covered in the procedural section (clause § E.11.5 "Procedures"/H.248.1).
regards Albrecht
Christian Groves <christian.groves@er To: Puglisi
Alberto
Alberto.Puglisi@italtel.it icsson.com> cc: megaco@ietf..org Sent by: Subject: Re: [Megaco] Call Flow Example in H.248.1 Corrigendum1 (03/2004) megaco-bounces@ietf. org
12.05.2005 05:47Hello Puglisi,
I'm not sure if I understand your question correctly. nt/dur shouldn't
have
the same value as nt/os, nt/or because Duration is in Milliseconds and Octets Sent and Received are in numbers of Octets.
Regards, Christian
Puglisi Alberto wrote:
- The MGC now sends both MGs a Subtract to take down the call. Only
the subtracts to MG2 are shown here. Each termination has its own set
of
statistics that it gathers. An MGC may not need to request both to be returned. A5555 is a physical termination, and A5556 is an RTP termination.
From MGC to MG2:
MEGACO/1 2 [123.123.123.4]:55555 Transaction = 50009 { Context = 5000 { Subtract = A5555 {Audit{Statistics}}, Subtract = A5556 {Audit{Statistics}} } }
From MG2 to MGC:
MEGACO/1 2 [125.125.125.111]:55555 Reply = 50009 { Context = 5000 { Subtract = A5555 { Statistics { nt/os=45123, ; Octets Sent nt/or=45123, ; Octets Received nt/dur=40000 ; in milliseconds } }, Subtract = A5556 { Statistics { rtp/ps=1245, ; packets sent nt/os=62345, ; octets sent rtp/pr=780, ; packets received nt/or=45123, ; octets received rtp/pl=10, ; % packets lost rtp/jit=27, rtp/delay=48, ; average latency nt/dur=38000 ; in millisec } } } }
I understand that nt/os, nt/or, and nt/dur should have the same
values,
am I correct ?
------------------------------------------------------------------------
Megaco mailing list Megaco@ietf.org https://www1.ietf.org/mailman/listinfo/megaco
Megaco mailing list Megaco@ietf.org https://www1.ietf.org/mailman/listinfo/megaco
Megaco mailing list Megaco@ietf.org https://www1.ietf.org/mailman/listinfo/megaco
_______________________________________________ Megaco mailing list Megaco@ietf.org https://www1.ietf.org/mailman/listinfo/megaco
_______________________________________________ Megaco mailing list Megaco@ietf.org https://www1.ietf.org/mailman/listinfo/megaco
participants (1)
-
Albrecht.Schwarz＠alcatel.de