RE: [AVT] RE: draft-ietf-avt-app-rtp-keepalive-01

Gunnar Hellström <gunnar.hellstrom@omnitor.se> Sat, 26 January 2008 15:37 UTC

Return-path: <avt-bounces@ietf.org>
Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1JIn6G-0003u0-9W; Sat, 26 Jan 2008 10:37:44 -0500
Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1JIn6E-0003tq-R7 for avt@ietf.org; Sat, 26 Jan 2008 10:37:42 -0500
Received: from s87.loopia.se ([194.9.94.113]) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1JIn6D-0006Rb-Jl for avt@ietf.org; Sat, 26 Jan 2008 10:37:42 -0500
Received: (qmail 77427 invoked from network); 26 Jan 2008 15:37:39 -0000
Received: from s34.loopia.se (HELO s60.loopia.se) ([194.9.94.70]) (envelope-sender <gunnar.hellstrom@omnitor.se>) by s87.loopia.se (qmail-ldap-1.03) with AES256-SHA encrypted SMTP for <avt@ietf.org>; 26 Jan 2008 15:37:39 -0000
Received: (qmail 48289 invoked from network); 26 Jan 2008 15:37:39 -0000
Received: from h240n1fls34o265.telia.com (HELO GunnarH) (gunnar.hellstrom@omnitor.se@[213.64.232.240]) (envelope-sender <gunnar.hellstrom@omnitor.se>) by s60.loopia.se (qmail-ldap-1.03) with SMTP for <avt@ietf.org>; 26 Jan 2008 15:37:39 -0000
From: Gunnar Hellström <gunnar.hellstrom@omnitor.se>
To: 'Randell Jesup' <rjesup@wgate.com>
References: <47679B43.3010503@db.org><DD8B8FEBBFAF9E488F63FF0F1A69EDD10441071E@ftrdmel1><E8308D99-774B-4B00-A5D8-72A9EFB97FDE@csperkins.org><DD8B8FEBBFAF9E488F63FF0F1A69EDD104471489@ftrdmel1><038901c85fef$d5e95400$0f75a7d9@GunnarH> <ybuhch0zm8y.fsf@jesup.eng.wgate.com>
Subject: RE: [AVT] RE: draft-ietf-avt-app-rtp-keepalive-01
Date: Sat, 26 Jan 2008 16:37:39 +0100
Message-ID: <03ae01c86031$627c33b0$0f75a7d9@GunnarH>
MIME-Version: 1.0
Content-Type: text/plain; charset="iso-8859-1"
Content-Transfer-Encoding: quoted-printable
X-Mailer: Microsoft Office Outlook 11
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2900.2962
Thread-Index: AchgKQ0dEHRBLqrURQSau7ZGaBHlQwABSNpg
In-Reply-To: <ybuhch0zm8y.fsf@jesup.eng.wgate.com>
X-Spam-Score: 0.0 (/)
X-Scan-Signature: a1852b4f554b02e7e4548cc7928acc1f
Cc: 'MARJOU Xavier RD-CORE-LAN' <xavier.marjou@orange-ftgroup.com>, 'Colin Perkins' <csp@csperkins.org>, 'AVT WG' <avt@ietf.org>
X-BeenThere: avt@ietf.org
X-Mailman-Version: 2.1.5
Precedence: list
List-Id: Audio/Video Transport Working Group <avt.ietf.org>
List-Unsubscribe: <https://www1.ietf.org/mailman/listinfo/avt>, <mailto:avt-request@ietf.org?subject=unsubscribe>
List-Post: <mailto:avt@ietf.org>
List-Help: <mailto:avt-request@ietf.org?subject=help>
List-Subscribe: <https://www1.ietf.org/mailman/listinfo/avt>, <mailto:avt-request@ietf.org?subject=subscribe>
Errors-To: avt-bounces@ietf.org

Randell,
Thanks for good comments.
You may be right that sending keep-alive with another SSRC may cause
unwanted effects in some receivers. But with an undefined PT as recommended
in the keepalive draft, I do not see higher risks than with sending it with
the same SSRC as another stream. 

About the requirement of text/red and text/t140 to be the only payload in
the stream, RFC 4103 says in section 5.3:
-----------------------------------------------------------------------
"The text/t140 format relies on use of the sequence number in the RTP
   packet header for detection of loss and, therefore, is not suitable
   for applications where it needs to be alternating with other payloads
   in the same RTP stream.  It would be complicated and unreliable to
   try to detect loss of data at the edges of the shifts between t140
   text and other stream contents.  Therefore, text/t140 is RECOMMENDED
   to be the only payload type in the RTP stream."
------------------------------------------------------------------
So, it is clear that the method proposed in 4.8 of the keepalive draft is
not directly suitable for RFC 4103. 

It would be good to define a method for RFC 4103 keepalive in the keepalive
draft.

Repetition of the latest transmitted RTP packet without modifying anything
could be another suitable method for RFC 4103 keepalive. Repeating also the
sequence number and time stamp, it will just be handled as a very late
duplicate and discarded by the receiver. Right?

About multi-party for RFC 4103. Yes, it is not explicitly defined in RFC
4103. Generla RTP methods can be applied.
But to be clear, multi-party session handling of RFC 4103 real-time text is
now described in 
http://www.ietf.org/internet-drafts/draft-hellstrom-text-conference-00.txt
 
I think SSRC collissions and allocating SSRCs etc is described sufficiently
in RTP RFC 3550, so I do not see a need for describing that more. Agree?

Thanks
Gunnar

-----Original Message-----
From: Randell Jesup [mailto:rjesup@wgate.com] 
Sent: Saturday, January 26, 2008 3:37 PM
To: Gunnar Hellström
Cc: 'MARJOU Xavier RD-CORE-LAN'; 'Colin Perkins'; 'AVT WG'
Subject: Re: [AVT] RE: draft-ietf-avt-app-rtp-keepalive-01

Gunnar Hellström <gunnar.hellstrom@omnitor.se> writes:
>The text/t140 and text/red payloads described in RFC 4103 rely on 
>inspecting the RTP sequence numbers for detection and repair of loss.  
>The sequence numbers roll on in one sequence for all payload types per 
>RTP stream.  Inserting other payload types in the same stream will 
>disturb the loss detection.  It will not be possible to detect what 
>payload type was lost.  This is noted in RFC 4103.

RFC 4103 appears to have an implicit assumption that the RTP stream is
unicast with only two participants.  There's no mention of 'unicast',
'multicast', multiple participants (modulo one sideways reference under the
congestion).  There's no mention of SSRC except in the format layout
diagrams.  Lack of consideration/discussion of these issues in the RFC is
part of the problem.  It does state that it expects related "multimedia"
streams to be in other RTP sessions.

I should note that because of how RFC 4103 is written and the redundancy
mechanism, it would be easy to define an RFC 4103-specific keepalive,
obviating the need to use this draft with it.  Or language could be inserted
in this draft specifying how to do it for RFC 4103.

>Therefore, the solution I see is to send the keep-alive RTP packets as 
>a separate stream within the same RTP session, thus using separate SSRC 
>from the main stream.

Effectively this mimics another sender within the RTP stream, a case which
is never explicitly discussed in the RFC.  This *should* be handled in the
normal RFC 3550 manner, but I'm mildly concerned that given the implicit
assumptions above, some implementations may not have considered the
possibility of multiple senders.

However, there's not much that can be done in that case anyways.

>I suggest to add the following wording of section 4.8:
>"The packet should also have an SSRC unique for the  keep-alive 
>traffic."

This needs to be more explicit about how the SSRC is chosen and dealt with
(collisions), and how this interacts with RTCP.  (Should there be a sender
report?)  Receivers might not have considered receiving reports for multiple
senders; again there may not anything that can be done.

Or, as above, add an RFC 4103-specific machanism.


--
Randell Jesup, Worldgate (developers of the Ojo videophone), ex-Amiga OS
team rjesup@wgate.com "The fetters imposed on liberty at home have ever been
forged out of the weapons provided for defence against real, pretended, or
imaginary dangers from abroad."
		- James Madison, 4th US president (1751-1836)


_______________________________________________
Audio/Video Transport Working Group
avt@ietf.org
https://www1.ietf.org/mailman/listinfo/avt

__________ NOD32 2824 (20080126) Information __________

Detta meddelande är genomsökt av NOD32 Antivirus.
http://www.nod32.com



_______________________________________________
Audio/Video Transport Working Group
avt@ietf.org
https://www1.ietf.org/mailman/listinfo/avt