Ken Mink wrote:
As a follow up to my earlier posts, I set up OpenVPN and my remote user is
using Xten's Eyebeam to connect to our * machine. Everything is connecting
up fine.
Great!
However, the voice quality is horrible. We've done some testing. If he comes
in via the vpn, the quality sucks. If I set up port forwarding and he comes
in that way, the voice quality is good, on par with a cell phone. The
network path is the same either way. The firewall machine is OLD, a p5-200.
The question is, do VPNs introduce latency that degrades the voice quality
by their very nature or is our vpn server so slow that the slow
encrypting/decrypting is causing it?
Very possibly. I see that Jon's already summarized this stuff well
between when i started this message, got pulled away, and now that I've
come back to finish. I'll stop babbling now. :)
The second question is what is the best
codec to use? It's currently set to gsm just because that's what it
defaulted to. Is there a better choice?
"better" choice is a very situational answer. You might get better
theoretical voice quality, at the expense of using more bandwidth (and
thus more encryption overhead) by using G.711. The increased overhead
of bandwidth and encryption might actually make things worse, though.
If you have a license for it on both ends, you might also consider
G.729, which both sounds good and is low bandwidth (on par with GSM).
The Asterisk server license for a single G.729 license will run you
about $10, and I think the Xten Pro software includes a G.729 codec
(again, ~$30 I think?). You need to evaluate the bandwidth available,
vs the quality you need, vs the amount of $$ you're willing to spend for
that quality, to come up with the "best" choice. Googling for 'voice
codec comparison' or something like "gsm g.729 g.711" will net lots of
descriptions about bandwidth usage for each codec. A good rule of thumb
is that G.729 is ~8kbps, GSM is ~13kbps, and G.711 is 64kbps. Keep in
mind those numbers don't include the packet headers, and there is a
packet sent every 20ms or 40ms with most setups, so the header sizes
*do* add up. Allow about 30 to 50% overhead above the bandwidth numbers
I just gave.
Aaron S. Joyner
--
TriLUG mailing list : http://www.trilug.org/mailman/listinfo/trilug
TriLUG Organizational FAQ : http://trilug.org/faq/
TriLUG Member Services FAQ : http://members.trilug.org/services_faq/
