One last note, given the Linux focus this thread has turned into: CF5 on the same physical 2-processor machine, in a tuned Linux RedHat 7 configuration just slightly outperforms CF5 on the same machine (swap drives) running Windows NT 4.0 AND Windows 2000 SP2.
CF5 also outperforms CF451 SP2 on the same SMP Linux RH7machine by nearly a factor of 2. -----Original Message----- From: Damon Cooper Sent: Wednesday, April 24, 2002 11:47 PM To: '[EMAIL PROTECTED]' Subject: Re: CF5 and SMP Performance...Tuning Best Practice The best practices formula is to load test to simulate your expected or observed load levels, in a controlled environment, while measuring avg page response time and throughput. Change one "tunable" knob at a time and re-test & measure until the optimal value for the tunable. Repeat until the optimal value is achieved. In the case of CF, the Sim Requests setting knob is the one that can normally yield the most significant results. Other knobs will be more or less important, depending on the app. In general, a good starting point for the Sim Requests setting tuning process is 3-4 requests/processor (was 2-3 for CF45 and below). Optimal value for your app will vary, of course. Some CF5 vs CF45 test results can be found here: http://www.macromedia.com/software/coldfusion/productinfo/performance_brief/cf5_perf_brief.pdf Regards, Damon ==================== Date: Wed, 24 Apr 2002 18:19:59 +0700 From: [EMAIL PROTECTED] (Paul Hastings) Subject: Re: CF5 and SMP Performance... Message-ID: <000f01c1eb81$f893f840$[EMAIL PROTECTED]> ...and the best practices formula for that would be? for nt4? for win2k? for linux? for unix? FAQ: http://www.thenetprofits.co.uk/coldfusion/faq Archives: http://www.mail-archive.com/[email protected]/ Unsubscribe: http://www.houseoffusion.com/index.cfm?sidebar=lists
