>I am not configuring any iteration at the moment so I think Cache manager will not help at the moment to my test case. Which isnt a realistic scenario and still doesnt account for shared JS/CSS/images that different requests will have in common for the same user
Id just disable the download embedded resources and see what values I get for different loads. How long does it take to get the page (avg in seconds) If the above value is large , then you have problems in your app. If the above value is small (and with embedded resources is large) then you have to perform the normal static optimizations (CDN, gzip, minify, collapse multiple files into 1, cache headers) that you can find using tools like YSlow , pagespeed etc. Also hopefully you are testing against a tuned (and realistic hardware) system, not some dev machine. regards deepak On Tue, Nov 16, 2010 at 3:09 AM, Amit <[email protected]> wrote: > Hello, > Thank you for your reply. > I am not configuring any iteration at the moment so I think Cache manager > will > not help at the moment to my test case. > Additionally we are trying to find out the maximum possible load our portal > can > take and hence trying to increase the user base step by step. > We have run upto 60 users with a ramp-up of Case 1 - 60 seconds and Case 2- > 30 > seconds ( 1 Thread per second and 2 Threads per second respectively). This > too > gave us the throughput of ~11 per min. > Can you please kindly advise how do we generate the load ? Is our method of > generating the load mentioned above is correct? > > regards, > Amit > > > > > > > > ________________________________ > From: Deepak Shetty <[email protected]> > To: JMeter Users List <[email protected]> > Sent: Tue, November 16, 2010 2:49:29 PM > Subject: Re: JMeter - Retrieve All Embedded Resource From HTML files > > >shall we consider 55 requests sent to the server ( 5 Threads * > >11(Number of subrequests)) and use this number to calculate the > throughput > and > >Avg. res time? > Usually no. The request normally refers to the number of pages a user sees. > However since the Request Embedded downloads resources serially without the > use of a cache , the throughput values will be worse than it actually is. > Adding a CacheManager should improve your values (and is more realistic) . > > There are other threads in this forum where felix, sebb and others have > provided their input , go through them.. > > >otherwise it hardly reaches 14 requests per minute. > For your example you havent really generated load , throughput values > should > increase with greater number of threads till it plateaus out and finally > falls. > > regards > deepak > > > On Mon, Nov 15, 2010 at 11:18 PM, Amit <[email protected]> wrote: > > > Hello, > > There is an option "Retrieve All Embedded Resource From HTML files" when > > configuring "HTTP Request HTTP Client". > > > > Now in my case, one page request corresponds to ~11 requests sent to the > > server > > for fulfilling the CSS,JS and images required for that page. > > > > > > In case of load testing the page let's say with 5 Threads and Ramp up of > > 2.5 > > seconds, shall we consider 55 requests sent to the server ( 5 Threads * > > 11(Number of subrequests)) and use this number to calculate the > throughput > > and > > Avg. res time? If I consider 55 request throughput comes in acceptable > > limits > > and otherwise it hardly reaches 14 requests per minute. > > > > > > At the moment JMeter shows the Avg. Res Time and calculates the > throughput > > based > > on number of threads. > > > > > > The site is hosted on Apache WS 2.2.17 and developed in PHP. > > > > Appreciate your insights and thoughts on this. > > > > Regards, > > Amit > > > > > > > > > > > > >
