>Pls correct my chain of events if it is wrong. No if you are dealing with logs then thats all youll check.
>Also, how do you work with such massive files Custom tool or log readers exist (which you can even load into DB) Again I dont use log files for perf testing so Im probably not the person to ask. >So if I don't have AppServer, then I don't have reverse proxy logs....correct? Basically this is whether you have a system that proxies requests to other systems (in the MS world this used to be MS Proxy Server , then ISA server). A load balancer is also a reverse proxy(with additional features and in some cases both are combined). >Is it the difference between DOMContentLoaded and Load time shown by Firebug? Yes that is the closest approximation. But in an AJAX world things get murkier!. As before accurate times are not needed and the amount of time a browser takes to just render doesnt depend on load so its a one time calculation. >Response for static files would increase respectively with the load. What is the point of measuring it if the datawould be irrelevant when I >apply the load. Because you need some estimate. As before you normally just need to know for e.g. pages should return under 10 sec for 30 concurrent users. Let say the time for 1 user is 1 second for the main page and 0.5 seconds for the static images/css/js etc. Lets say under 30 concurrent users (and usually you might take readings for 10, 20 users so you know your trend) Case a - The time for the main page is under 4 seconds - You know that dynamic code hasnt worsened much so static code wont either Case b. - The time for the main page is about 5-6 seconds - You know this is bad , you'd probably tune. You might try to see what happens with static but mostly you should be under your goal of 10 seconds (but close to it so thats not good!) Case c - The time for the main page is 8+ seconds - This looks to be increasing exponentially. System probably cant handle 30 users concurrently (So static values are irrelevant) Static files will always behave better than dynamic files and if your dynamic files behave well (under load) then static will behave well. If your dynamic files dont behave well then you have a problem anyway and the static numbers are irrelevant . The other thing is tuning for static files is pretty well defined so most applications can always pretty much optimize their static files and no improvement is possible there. And finally you will have limited time , so use it for stuff that matters. Static files rarely matter (subject to usual caveats and stuff). regards deepak On Tue, Sep 21, 2010 at 11:06 AM, Prostak <[email protected]> wrote: > > >Your IIS would have logs which > >state when the request started and ended as should your Load Balancer. > > These files are massive. So a tester needs to find a time for a single > request in LB logs => then find a time for the same request in server logs > => then find a time for the response => then find the time for the same > response in LB logs. > Then he should subtract all the times to get an idea of a response times. > Pls correct my chain of events if it is wrong. > > Also, how do you work with such massive files? Just simple "Find" method in > notepad (for windows based)? > > >Java based applications have a WebServer which proxies requests to the > >actual application servers. > > So if I don't have AppServer, then I don't have reverse proxy > logs....correct? > > >You can figure out how much > >time representative browsers take to render your page. > > Is it the difference between DOMContentLoaded and Load time shown by > Firebug? If no, how do you measure rendering time? > > >You can see from a > >browser what time is needed for static files download (cached and non > >cached). > > Here is where I am puzzled. Remember that my website is on a single Web > Server (no Appl Servers). Response for static files would increase > respectively with the load. What is the point of measuring it if the data > would be irrelevant when I apply the load. So if images (etc) take 2 > seconds > to load, I can't use this time to add because it will increase when the > load > is applied. > -- > View this message in context: > http://jmeter.512774.n5.nabble.com/Test-plan-for-970-page-requests-every-5-min-tp2826174p2848514.html > Sent from the JMeter - User mailing list archive at Nabble.com. > > --------------------------------------------------------------------- > To unsubscribe, e-mail: [email protected] > For additional commands, e-mail: [email protected] > >
