> Anyway, I'm almost sure that GnuGK will crash with 10 Setup/sec. This way > you'll have 20E1 in ONE minite. We may consider it the top level for full > proxy installs. To keep the number of sim. calls in this level we need 10 > ReleaseComplete/sec. So, ACD will be < 60 sec. which may occur only if you > are the last priority (e.g. 15th) gateway for a given destination. > > Another thing is that processing large socket select lists is not efficient > > - it is probably better to split them among multiple threads. > Won't context switching be more expensive then that socket list processing? That's interesting - it would be great if someone could do some test to compare different CallSignalHandler settings. We could introduce a few measures like ASR,PDD, jitter and average audio delay. But I can imagine that average dialing phase (from setup to altering - PDD) would be longer with CallSignalHandlerNumber=1. Another situation when single handler will be a bad choice is when accounting/authentication takes long time (due to unresponsive backend server or some other issues). Anyway, if single threaded design would offer enough performance, there would be no need to create the gk 2.2... --- Michal Zygmuntowicz ------------------------------------------------------- This SF.net email is sponsored by: SF.net Giveback Program. Does SourceForge.net help you be more productive? Does it help you create better code? SHARE THE LOVE, and help us help YOU! Click Here: http://sourceforge.net/donate/ _______________________________________________ List: Openh323gk-users@lists.sourceforge.net Archive: http://sourceforge.net/mailarchive/forum.php?forum_id=8549 Homepage: http://www.gnugk.org/
