>I'm not sure why Deepak doesn't like this feature. The problem as I see it is that for static files , the client network , use of cache , the browser you are using, whether the static content is served from a CDN or not is significant (because the actual time to serve these files is really small). Also since the webserver is fairly constant with respect to the response times for the loads I use (because the app server will fall over much before the webserver does) I can estimate this for different values of network / browser / new-returning user without needing to run a test.
Or lets put it another way if Chrome takes 3 seconds to render a page whereas IE6 takes 4 seconds and Jmeter without embedded resources is 2.5 seconds for the same page what value do you expect Jmeter with embedded resources to return to you and what is your experimental error %? So if accurate response times are needed , I cant use this feature and if accurate response times are not needed then I dont need this feature anyway. >You should mess around with the different options to find out what works >best for you. Sound advice. regards deepak On Tue, Sep 14, 2010 at 4:11 AM, Felix Frank <[email protected]> wrote: > > This is an interesting topic and your answers are really informative. > > Why do you recommand using Firebug? Is Firebug not loading static files > and > > apsx files simultaneously? > > Firebug's Net tab will illustrate Firefox's behaviour wrt. parallel > downloads. You should use it to get an idea of how typical (non-Jmeter) > client interaction works with your application. > > > An other question I have is: Do you have 'Retrieve All embedded reourses' > > enabled or disabled when loadtesting a Webserver? I couldn't really make > > that up from your previous posts in this topic. > > I'm not sure why Deepak doesn't like this feature. My personal rule of > thumb is to always use it (in conjunction with the Cache Manager) to do > what the browser does while > a) using as few Samplers as possible and > b) allowing Jmeter to compute average times based on times to load whole > pages, not each tiny image under test > > Do note that there are caveats to this approach as well. Failure to load > any embedded image or script will count as a page fault in your > statistics and other effects. > > You should mess around with the different options to find out what works > best for you. > > Cheers, > Felix > > --------------------------------------------------------------------- > To unsubscribe, e-mail: [email protected] > For additional commands, e-mail: [email protected] > >
