[asterisk-dev] Native Bridging: terminology

[Mark Michelson]

Klaus Darilion wrote:
> Hi!
> 
> What is the exact meaning of "native bridging"? For example I know with 
> DAHDI it means direct bridging on the hardware. But what about from one 
> DAHDI interface card to another DAHDI card? Are there some optimizations 
> that the audio stays within the kernel module or is it always send via 
> Asterisk's core?
> 
> Or in case of a SIP-to-SIP call with canreinvite=no. Will there be some 
> optimizations that audio stays within the rtp module or will it be sent 
> via Asterisk's core? If there are such optimizations - is this called 
> direct bridging or is direct bridging the RTP exchange between clients 
> directly using canreinvite=yes (which I would not call direct bridging 
> but "offload" or "bypass")? Or is this the "external native bridge"?
> 
> thanks
> klaus
> 

The term "native bridge" is a bit overloaded in Asterisk. On the surface, it 
simply means that the two legs of a call have the same channel->tech->bridge 
function.

I can't speak for DAHDI, but within SIP, native bridging has two subcategories. 
One, typically referred to as "SIP native bridging" is used when reINVITEs are 
enabled. The endpoints send their media directly to one another. The other 
subcategory is called "Packet 2 Packet" or "P2P" bridging. If reINVITEs are not 
enabled, but there are also no features that require the Asterisk core to be in 
the voice path, then the bridging will be done at the RTP layer of Asterisk.

I hope that helps.
Mark Michelson