Hi,
I'm working on Project Clearwater (http://www.projectclearwater.org/), an open
source highly-scalable IMS (IP Multimedia Subsystem) implementation.
We've been using SIPp for stress testing. Since we use SIP/TCP and scale up to
10s of millions of subscribers (and hence 10s of millions of TCP connections),
we've had to make a few enhancements - most notably to use Linux's epoll
support for polling and asynchronous connects. These are in the epoll branch
(http://sourceforge.net/p/sipp/git/ci/epoll/tree/) and hopefully will be merged
into master soon. With these changes, we've successfully got 50000 TCP
connections from a single instance of SIPp (and the limitation we then hit is
with Amazon AWS EC2, on which we're running, not SIPp itself).
We've also tried running the IMS Bench SIPp benchmark against Clearwater. We
were surprised to find that this only uses a single pair of TCP connections
(possibly a limitation of SIPp's call routing model), which seems quite
unrealistic. We want to be able to use IMS bench to test our P-CSCF node,
where we'd expect to see one TCP connection per endpoint, rather than muxing
multiple endpoints down a single connection. Unfortunately, we couldn't
convince IMS bench to run in TCP multi socket ("-t tn") mode. Does anyone have
any experience running IMS Bench SIPp at high scale over TCP or to test access
devices? Please let me know if so.
Thanks,
Matt
------------------------------------------------------------------------------
See everything from the browser to the database with AppDynamics
Get end-to-end visibility with application monitoring from AppDynamics
Isolate bottlenecks and diagnose root cause in seconds.
Start your free trial of AppDynamics Pro today!
http://pubads.g.doubleclick.net/gampad/clk?id=48808831&iu=/4140/ostg.clktrk
_______________________________________________
Sipp-users mailing list
Sipp-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/sipp-users
