OK, now everything goes back to the original server. We have reached an acceptable performance using cx_Oracle instead of DCOracle2. It seems that our DCOracle2 module was not behaving multithreaded (anybody translate that into English). Only one query was active at a time.
We are getting 8 pages/second/Webware process (Python only uses 1 cpu). El lun, 10-03-2003 a las 21:12, Marcos Sánchez Provencio escribió: > Not really... > > We have put Pound+2xApache+2xWebware in a separate linux machine. Now, > with 400 virtual clients (using an automated testing tool) the linux > machine is basically idle (each one of 4 CPUs 90% idle). The solaris > machine that hosts the oracle goes to 90+% on each of 4 CPUs, but > contonues to give fast service to the vintage cobol app that we are > partially replacing (I mean, it accesses about the same data as the > webware app). > > So, what is the recommended setup (apache+webware+DBPool) for high > performance/high load? I expect to get about 1000 users asking for 5 > pages per minute (each page takes about 6 oracle queries). Are there any > records of this kind of load? With what hardware? > > I have a variation on the DB pool that comes with webware (I will post > it tomorrow at work). I am using DCOracle2. Could cxOracle be better? > SQL Relay? Any alternatives to test? Are there any caveats about using > threads in Webware? with DCOracle2 accesing Ora7? > > One strange thing. We have set up Webware for 100 threads on each > instance max, but they never get started. If I access the example pages > (wich obviously don't access Oracle) they take forever to load (minutes) > although we don't get the thread max. The apache sample pages load > immediatly. > > My objective is to be able to say that Oracle is the next bottleneck, of > course. > > I am sorry to be so boring, but I feel this is going to determine the > future of Webware both in my company and in the Spanish Public > Administration. > > Thank you for your time in reading this legthy mail. > > El lun, 03 de 03 de 2003 a las 20:00, Edmund Lian escribió: > > On 03/03/2003 12:17:32 PM Marcos wrote: > > > > >Given an average load (100 users) it is the apache-ssl process that's > > >eating the 4 CPUs. Our next task is lightening the served pages, in hope > > >that this will lighten the burden of encrypting the traffic. On the > > >other side, I expect the webware process to be the next bottleneck :-) > > > > Hey! You leave poor Webware alone--your next bottle neck is going to be > > Oracle, OK? :-) > > > > In Webware we trust. :-) > > > > ...Edmund. > > > > > > > > ------------------------------------------------------- > > This sf.net email is sponsored by:ThinkGeek > > Welcome to geek heaven. > > http://thinkgeek.com/sf > > _______________________________________________ > > Webware-devel mailing list > > [EMAIL PROTECTED] > > https://lists.sourceforge.net/lists/listinfo/webware-devel > -- > Marcos Sánchez Provencio <[EMAIL PROTECTED]> ------------------------------------------------------- This SF.net email is sponsored by:Crypto Challenge is now open! Get cracking and register here for some mind boggling fun and the chance of winning an Apple iPod: http://ads.sourceforge.net/cgi-bin/redirect.pl?thaw0031en _______________________________________________ Webware-devel mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/webware-devel
