Re: jain-sip (nist) performance
Man-Chi Leung <[email protected]>
| Newsgroups | gmane.comp.voip.nist-sip,gmane.comp.java.com4j.user |
|---|---|
| Message-ID | <[email protected]> |
hi, sorry to trouble u, I would like to understand more about performance issue on SIP. it is not uncommon to see commerical stacks claiming to be "hi performance" sip stack. in convention, how hi is considered as hi performance ? 50CPS? 100CPS? or x00CPS? :-) it there any guideline for this performance benchmarking? any advice and area that we need to aware during performance testing on SIP (e.g. dialog, session) ? sorry for my ignorance.. pls advise ~manchi On 2006-01-24 19:43:35 +0800, "M. Ranganathan" <[email protected]> said: > Hello! > > Some folks have reported 50 CPS ( on a sun netra 4 processor machine > using a jsip Resource Adaptor built on jsip 1.2 ). > > 1. Do not use dialogs (javax.sip. AUTOMATIC_DIALOG_SUPPORT=off in the > configuration properties of the stack). > 2. Use JAIN-SIP-1.2 (please get it from CVS). > > However, you should not expect amazing performance. If you are looking > for amazingly good performance, perhaps you can try a commercial vendor > for jsip. Post on jain-sip-interest and perhaps you will get a > response. I am not allowed to make recommendations. > > Do not treat the proxy on java.net as your yardstick for performance. > It does not perform very well and besides has not been upgraded to > jsip1.2 yet. > > > Regards > > Ranga > > Francisco Javier Tellez Vaquero wrote: > >> Hi all, >> we've made several performance tests against jain-sip(nist) stack with >> SIPP as the client and no more than 14 calls per second are obtained in >> a stable way. >> >> The workflow was: >> - Send INVITE >> - Receive 180 INVITE >> - Receive 200 INVITE >> - Send ACK >> - Send BYE >> - Receive 200 BYE >> >> We used good machines with several CPU's, more than 4Gb of memory, ... >> I'd like to know if there's a way to increase performance or it's the >> maximum we can obtain. >> >> Regards, >> Fran. >> >> >> --------------------------------------------------------------------- >> To unsubscribe, e-mail: [email protected] >> For additional commands, e-mail: [email protected]