I think your methodology is sound if a single variable is in isolation. The variable "Simultaneous requests" is not in isolation however. There are plenty of other tunable variables that can affect what value is best for simultaneous requests. The only way to properly tune a system and application is understand how each variable can affect the others. There are simply too many permutations to test every possible variable. Thus you need someone who understands the OS as well as someone who understands the application. Since what is tunable varies greatly between OSs, the methodology taken is in fact different between OSs.
> We seem to be arguing past each other a lot lately, Matt. I hope you're > having fun! > I'll never pass up a good debate. It is the only way to learn! -Matt > -----Original Message----- > From: Dave Watts [mailto:[EMAIL PROTECTED]] > Sent: Wednesday, April 24, 2002 1:23 PM > To: CF-Talk > Subject: RE: CF5 and SMP Performance... > > > Except that there is a significant number of values to look > > at, not just some CF setting. What these values are heavily > > depends on the OS system. Understanding the difference in > > threading models allows one to tune these values more > > appropriately. > > So, what would you do differently, given two boxes to tune, let's say > Solaris and Linux? How would you find the optimal value for simultaneous > requests on each? How would you know that you'd found the optimal value in > each case? > > Given one specific tunable variable, the process for determining the > optimal > value for that variable is the same, using a load-testing methodology, > without regard to what that variable is. Yes, there are many tunable > variables. Yes, knowing what they represent will help narrow the scope of > necessary testing. Yes, the process will move faster if you know what > tunable variables are most likely to require and benefit from tuning. That > doesn't mean that the process itself will be any different between > platforms, which again was my point, to which you objected. > > However, the point of using an external methodology is that it allows you > to > apply statistical analysis instead of expertise and knowledge of best > practices. This testing methodology can be implemented by competent QA > engineers, rather than by expert developers and sysadmins. If used in > conjunction with a competent sysadmin, it can find which server tuning > parameters willl make the biggest difference in performance. If used in > conjunction with expert developers, it can find the areas within the > application code that are most in need of attention, allowing the expert > developers to maximize the value of their expertise. It also provides full > coverage, in that you don't have to worry about your expert developers > missing something in their code review - it'll show up as a bottleneck > during testing. If used during the development process, this methodology > allows you to play some interesting "what if" games, which can save your > ass > before deployment. > > The problem with this methodology that I haven't addressed is cases of > interdependent values, which is the place where expertise is necessary, > but > the vast majority of cases don't fall into that category. You might be > surprised how much difference finding the optimal value for "simultaneous > requests" can make. I don't know about you, but that's something that I > can't determine solely through expertise. > > We seem to be arguing past each other a lot lately, Matt. I hope you're > having fun! > > Dave Watts, CTO, Fig Leaf Software > http://www.figleaf.com/ > voice: (202) 797-5496 > fax: (202) 797-5444 > > ______________________________________________________________________ This list and all House of Fusion resources hosted by CFHosting.com. The place for dependable ColdFusion Hosting. 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
