tag:blogger.com,1999:blog-7557978943964322249.post7382212993228579090..comments2008-05-02T22:43:24.440+03:00عادلadel.sh@gmail.comBlogger5125tag:blogger.com,1999:blog-7557978943964322249.post-58662842698705327202008-05-02T20:35:00.000+03:002008-05-02T20:35:00.000+03:00I guess most people assume that performance testing always means server performance, but I would like to do end-user performance testing.<BR/><BR/>Basically, I have a web-based application that will be used worldwide. I'm being asked to document how test scenarios will perform in 80-90 different locations -- each with unique network characteristics. From extrmely high latency to very constrained bandwidth. Some as poor as <160KBs and 800+ms.<BR/><BR/>So, what I want is a single test client that can execute test scenarios that range from 20-50 page navigations or submissions. I want the entire functional scenario to be scripted and I want to re-configure my lab network simulation to cover each of the locations under test.<BR/><BR/>In this case, can Selenium give me the per page test results and overall scenario response times I need? Are there better ways to do this kind of testing?Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-7557978943964322249.post-15342102621228893352008-04-10T15:46:00.000+03:002008-04-10T15:46:00.000+03:00opsssst<BR/>it looks it's me who misread your fist comment. <BR/><BR/>What you say makes a lot of sense :)Test Automation Bloghttp://testsquad.orgnoreply@blogger.comtag:blogger.com,1999:blog-7557978943964322249.post-72592688225512476922008-04-10T13:00:00.000+03:002008-04-10T13:00:00.000+03:00The point I was trying make is that Selenium is a functional/Regression automated testing tool.<BR/><BR/>If you wanted, you could set up a few dozen machines with QTP and get a "performance" test going. The same could be said with Selenium and having a few dozen instances of the test running to get a performance test. It can be done but is it really worth it?<BR/><BR/>There have also been documented instances where Selenium doesn't scale well. In my blog post about what makes a good automated test there is a video attached. In there is a Google Farm Engineer complaining that Selenium doesn't scale well so can't be used as a performance tool.<BR/><BR/>Konstantin Petkov posted about how he tried to use it a performance tool (http://blogs.telerik.com/KonstantinPetkov/Posts/07-08-20/ASP_NET_Web_Control_Performance_Test_in_Selenium.aspx?ReturnURL=%2fKonstantinPetkov%2fPosts.aspx) and he agreed that it wasnt a good idea.<BR/><BR/>I hope that I have expanded on my argument a bit better this time :)David Burnshttp://www.blogger.com/profile/11233457684822187675noreply@blogger.comtag:blogger.com,1999:blog-7557978943964322249.post-10787653682949228292008-04-10T12:49:00.000+03:002008-04-10T12:49:00.000+03:00How do you use QTP for performance testing? <BR/><BR/>What i know is QTP is a functional testing automation tool. Maybe you mean LoadRunner or Jmeter? <BR/><BR/>AdelTest Automation Bloghttp://testsquad.orgnoreply@blogger.comtag:blogger.com,1999:blog-7557978943964322249.post-63178129459259091082008-04-10T00:16:00.000+03:002008-04-10T00:16:00.000+03:00Like Haw-bin said you can use Selenium to do performance testing. Unfortunately I would liken it to using QTP to do performance testing. There is a lot of overhead just to do a performance test!<BR/><BR/>Functional automated testing tools like Selenium and QTP are better used for functional and regression testing. If you read my blog posting on <A HREF="http://blog.theautomatedtester.co.uk/2008/03/what-makes-good-automated-test.html" REL="nofollow">what makes a good automated test</A> you can see that having the re-usability is worth a lot more than the performance test in this case.David Burnshttp://www.blogger.com/profile/11233457684822187675noreply@blogger.com