Thanks for the 'tips', I did add temporary delays in my voip pipe
definitions as you suggested, and I did see the expected increase in
my SIP 'pings'. So, my pipes are working.
I am using just pipes, no queues for voip... if I add a single queue
to each pipe and reference the queue with my rule, no noticeable
difference in the latency.
Interestingly, if I look an ICMP pings, the traffic shaper only adds
about 5ms round trip vs 40ms for a SIP 'ping', going through the same
pipes. There must be something about asterisk SIP 'pings'.
My ear can not tell the extra latency the traffic shaper is adding.
But, now that I know, it 'bugs' me... <smile>
On Aug 21, 2006, at 5:26 AM, David W. Hess wrote:
> I switched to pfsense about the time I started using VOIP but back
> in my gaming
> days I did a lot of work with m0n0wall to reduce latency for gaming
> The best configuration I found for latency was to use 2 pairs of
> pipes with the
> total throughput less then the DSL line throughput. One pair was
> for all non
> latency sensitive traffic while the other pair was used for VOIP,
> gaming, and
> DNS. With this setup I never noticed any significant change in
> latency with
> traffic shaping turned on or off and saturating the non latency
> pipes had only a
> very small effect on the latency pipes. I had one queue assigned
> to each
> latency sensitive pipe with all traffic assigned to that queue.
> How is Asterisk measuring latency? Is it possible that the
> measurement packets
> are not going through the same pipe as the VOIP traffic? You could
> verify this
> by using the delay option.
> On Sun, 20 Aug 2006 20:00:01 -0500, you wrote:
>> Continuing this discussion of VOIP and m0n0wall traffic shaping...
>> (m0n0wall on Soekris net4801)
>> Using asterisk, all my outbound SIP are set with "qualify=yes".
>> At the asterisk command line, a "sip show peers" shows the latency to
>> the peers.
>> I notice that with traffic shaping enabled, 40+ ms is added to the
>> peer latency in the "sip show peers", for all my SIP peers.
>> Question: Is this a true delay added by the shaper that adds to the
>> RTP traffic latency?
> To unsubscribe, e-mail: m0n0wall dash unsubscribe at lists dot m0n0 dot ch
> For additional commands, e-mail: m0n0wall dash help at lists dot m0n0 dot ch