Summary:ASTERISK-07777: Add rtp-packetization.txt to doc directory
Reporter:dea (dea)Labels:
Date Opened:2006-09-19 16:12:27Date Closed:2006-09-20 12:41:56
Versions:Frequency of
Environment:Attachments:( 0) packetization-docs-v3.diff
Description:Provides a brief overview of the framing options available in
chan_sip, chan_skinny and chan_ooh323.  

A table of codec framing parameters is included to help guide the
administrator to configuring valid framing options.

skinny.conf.sample and sip.cong.sample updated to refer to this document.
Comments:By: Brian West (bkw918) 2006-09-19 16:18:03

That max table seems too low.. You can clearly send 120ms g711 that will be an RTP packet of 960+12byte header which is very efficient :P.  The maxes should go up to the RTP MTU but not over.


By: dea (dea) 2006-09-19 16:28:28

The max values are likely on the conservative side, being culled from RFCs, Cisco whitepapers and anywhere that claimed knowledge.

The document reflects the values currently implimented.  Adjusting the implimented values is fodder for another discussion, but at least there
should be something to guide sysadmins how to use this feature.

By: Serge Vecher (serge-v) 2006-09-19 16:40:15

dan: a few spelling issues, if you will:

1) "Asterisk now support" -> "Asterisk currently supports"
2) "select RTP based" -> "select RTP-based"
3) "remote endpoints preferences" -> "remote endpoint's preferences"
4) "depends on the endpoints to present the desire packetization" -> "... endpoint's ability ... desired ..."
5) "example allow=ulaw:33 will set the coded to 30ms framing" -> "...codec..."

By: dea (dea) 2006-09-19 17:02:20

Comments addressed in packetization-docs-v2.diff, except that 'svn diff' caught
some spurious h323/Makefile changes.  

Ignore v2 and use v3 please.

By: Serge Vecher (serge-v) 2006-09-20 11:56:56

Dan, one more thing, the internal chan_h323 is packetization-capable as of r43331
this morning. Technically, chan_ooh323 in addons is not yet capable, as your patch in ASTERISK-5440 has not yet been merged in, but hopefully objsys will get to that quickly...

By: Brian West (bkw918) 2006-09-20 12:03:55

You should still note that those are not hard coded MAX values but recommended max values.  You have 230 for g729 which granted is a small packet vs a 120ms ulaw packet at 960bytes.  But still don't give people the false impression that they have some sort of limit when infact they don't.  It should be noted that those are recommend values and should not exceed the RTP MTU.


By: Jason Parker (jparker) 2006-09-20 12:41:55

Committed to svn trunk in revision 43344.