Hello Chris,
In the tests we performed we used SIPp as the UACs and the UAS was an
OpenSIPS. In spite of this, I doubt that the UAS SiPp in your scenario
would be the limiting factor.
You could try the load debugging patches available in trunk ( more info
on this can be found at [1] )
to see what exactly is the limiting factor. I do not think the problem
is the hardware you are testing on, as an dual core Intel with 4Gb of
RAM should support plenty of traffic.
As Adrian pointed out, it is much more likely that external queries,
like DNS, SQL or Radius are the bottleneck on your system. Make sure to
have OpenSIPS run in fork mode with enough children, and try the load
debugging features to pinpoint your problem.
[1] http://lists.opensips.org/pipermail/users/2011-February/016918.html
Regards,
--
Vlad Paiu
OpenSIPS Developer
On 03/09/2011 01:14 PM, Adrian Georgescu wrote:
The round trip time to the database and number of DNS, SQL and Radius queries
will impact the maximum possible CPS.
You have to factor them all in.
Adrian
On Mar 9, 2011, at 10:44 AM, Chris Martineau wrote:
Hi,
Don't understand how you get such high cps. I have a scenario using SIPp
as the clients and a script similar to test_I config file. The scenario
makes a call to the far SIPp client waits 10 seconds and then sends a
bye with the far SIPp client responding appropriately. However I can
only manage a maximum of around 30cps before the 100 trying messages
from the server start to fail and the invite retransmissions start to
cascade.
I have xlog disabled and debug=0 as this grinds the system down to only
5cps when switched on.
My setup is a dual core intel with 4gb.
As you can see there is a vast difference between 30 and 6000 cps which
I can't imagine is being that drastically effected by my hardware as
processor overhead is barely touched when it fails which would indicate
opensips is introducing delay by blocking somewhere!
The system is on a basic CentoS install.
By the way I have also tried this with a stripped down config similar to
test_B with little difference in performance!
Any suggestions?
Many thanks
Chris
-----Original Message-----
From: [email protected]
[mailto:[email protected]] On Behalf Of Vlad Paiu
Sent: 08 March 2011 14:16
To: OpenSIPS users mailling list; OpenSIPS devel mailling list
Subject: [OpenSIPS-Users] OpenSIPS performance tests
Hello,
We have performed some tests using OpenSIPS 1.6.4 to simulate some real
life scenarios, to get an idea on the load that can be handled by
OpenSIPS and to see what is the performance penalty you get when using
some OpenSIPS features like dialog support, DB authentication, DB
accounting, memory caching, etc.
A total of 11 tests were performed, using 11 different scenarios. The
goal was to achieve the highest possible CPS in the given scenario,
store load samples from the OpenSIPS proxy and then analyze the results.
The base test used was that of a simple stateful SIP proxy. Than we kept
adding features on top of this very basic configuration, features like
loose routing, dialog support, authentication and accounting.
The detailed description of the test scenarios performed, a performance
penalty table showing the load impact introduced by adding different
features onto OpenSIPS as well as a graphical chart showing load samples
from all the tests can be found at
http://www.opensips.org/Resources/StressTests
Regards,
--
Vlad Paiu
OpenSIPS Developer
_______________________________________________
Users mailing list
[email protected]
http://lists.opensips.org/cgi-bin/mailman/listinfo/users
_______________________________________________
Users mailing list
[email protected]
http://lists.opensips.org/cgi-bin/mailman/listinfo/users
_______________________________________________
Users mailing list
[email protected]
http://lists.opensips.org/cgi-bin/mailman/listinfo/users
_______________________________________________
Users mailing list
[email protected]
http://lists.opensips.org/cgi-bin/mailman/listinfo/users
