Maksym,
Thank you for your answer. I'll try it tonight when the load on the proxy is
low.
Could this also be the reason for the relative high number of 'packets to
unknown port received' when issuing a 'netstat -us':
netstat -us
Udp:
3590520622 packets received
3397618 packets to unknown port received.
4074 packet receive errors
3601109676 packets sent
This number is increasing relatively fast. I don't know how I can make sure
that these packets are RTP packets. I think that these packets will be dropped
(which CAN cause the jitter/dropped calls towards the ITSP).
Regards,
Ronald
-----Oorspronkelijk bericht-----
Van: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] Namens Maxim Sobolev
Verzonden: dinsdag 24 juni 2008 0:09
Aan: RTPproxy Users
Onderwerp: Re: [RTPproxy Users] Performance (Delay / Jitter)
Ronald Voermans wrote:
> Hello,
>
> We are using OpenSer with RTPProxy for serveral years now. We recently
> changed the hardware on which OpenSer/RTPProxy is running, and upgraded
> all the software.
>
> I'm using OpenSer 1.3 with RTPProxy 1.1.
>
> We are forwarding all PSTN request to our ITSP. We are doing max. 40
> simultaneous calls. Since a couple of weeks our customers are
> complaining that their calls are being dropped, and that there is a
> significant amount of jitter.
>
> I have been looking for the cause of this for some time now, and have
> come to the conclusion that it must be somewhere between the OpenSer and
> our ITSP (have done serval monitoring on the customer site, and at our
> site). Our ITSP says everything is running fine (as usual J), so I have
> to come up with some evidence that the cause is indeed with the ITSP.
>
> First thing I would like to do is check if I've optimized everyting on
> the (new) OpenSer server:
>
> The server is a HP Proliant DL140; a quadcore Intel processor (2.44
> GHz), with 4 GB of RAM. CPU is never above 0.1% and memory usage < 1 GB.
>
> I'm running only 1 instance of RTPProxy. In the FAQ's I read that you
> can better run 4 instances of RTPProxy. Is this also necessary with my
> setup, and with the max. number of sim. calls I have? Will this have a
> performance-gain if I run 4 instances? If so, could this be the cause of
> the jitter/dropped calls?
Roland,
I don't think there will be any improvements if you run several
instances of the proxy in parallel with this amount of traffic. What
could help a bit, though, is increasing size of kernel UDP buffers to
avoid packet loss during traffic spikes, on Linux you can use the
following commands:
sysctl -w net.core.rmem_max=16777216
sysctl -w net.core.wmem_max=16777216
Commands above allocate 16MB to the send buffer and 16MB to the receive
buffer. The default value is 128KB, which is too low for loss-sensitive
VoIP.
If it doesn't help, the way to investigate the issue is to capture
network dumps either on the machine running proxy or even better on
network's gateway. The tools like Wireshark (former Ethereal) could help
you to analyze RTP streams given these captures. You can for example
compare loss level and jitter for streams coming to the proxy and
streams going out and show this data it to your ITSP.
Just for your information we are currently working on feature that would
gather real-time RTP loss/jitter statistics in rtpproxy. It will
probably available in 2-3 weeks from now.
Sincerely,
--
Maksym Sobolyev
Sippy Software, Inc.
Internet Telephony (VoIP) Experts
T/F: +1-646-651-1110
Web: http://www.sippysoft.com
_______________________________________________
Users mailing list
Users@rtpproxy.org
http://lists.rtpproxy.org/mailman/listinfo/users
_______________________________________________
Users mailing list
Users@rtpproxy.org
http://lists.rtpproxy.org/mailman/listinfo/users
