[asterisk-dev] [Fwd: [svn-commits] oej: branch oej/iaxtrunkfix-1.2 r44779

[Johansson Olle E]

10 okt 2006 kl. 17.24 skrev Rich Adamson:

> Author: oej
> Date: Tue Oct 10 04:43:12 2006
> New Revision: 44779
>
> URL: http://svn.digium.com/view/asterisk?rev=44779&view=rev
> Log:
> Add patch for setting MTU on trunks.
>
> With a large amount of traffic on IAX2 trunks, there is bad voice  
> quality due to the fact that the
> IAX2 trunking scheme depends on the Linux system to handle  
> fragmentation of UDP packets. This is
> not very efficient. This patch adds the ability to set an MTU size  
> for *ALL* iAX2 trunks in
> Asterisk 1.2. In the patch for svn trunk, we will add new features,  
> like setting MTU per trunk.
> In the case of Asterisk 1.2, we wanted a small clean patch to be  
> able to get voice quality back
> without changing a lot of source code.
>
> --------------------------------------------------
>
> For those that get very serious about QoS in WAN and LAN links,  
> fragmenting packets to various sizes depending on actual bandwidth  
> available is very very important.
>
> Couple of examples:
> 1. On T1/E1/ethernet circuits (and faster), fragmentation is of  
> little concern as a 1500 byte packet consumes approx .9  
> milliseconds to transmit (across a T1). That suggests there are  
> many opportunities to squeeze in a 20 millisecond rtp packet.
>
> 2. On a sdsl circuit with 512k bandwidth in both directions, a 1500  
> byte packet will consume substantially more transmission time,  
> impacting the ability to squeeze a 20 millisecond rtp packet into  
> the serial stream.
>
> The issue that occurs in #2 is that once a large (1500 byte) packet  
> begins to flow across a slow speed link, there is no way to stop  
> that flow to insert a 20 milliscond rtp packet, thus creating  
> jitter at best. QoS settings (in any form of box) will not impact  
> this problem.
>
> Cisco has created methods in its latter IOS versions that forcibly  
> fragments large packets so that QoS has a reasonable chance of  
> moving rtp packets and minimizing jitter.
>
> The actual MTU size to be used should be a parameter of each  
> transmission path, and again, the MTU size "is" directly related to  
> available bandwidth.

As indicated in the commit comment, this will be possible with the  
patch for svn trunk. I don't want to add new features
to 1.2/1.4 beta.

Thanks for the feedback! Please test these branches and give feedback.

/O