I had thought that the current versions of Loadrunner were up on current web tech, but it's been years since I used it.
If you are running windows servers, then the loadtest tools that come with the 'ultimate' version of dev studio are generally the way to go. It does most of what loadrunner does (and more to some degree) in that environment and is far more cost effective. Great support too via the msdn forums. It's not practical to generate load using real browsers, it's also generally not necessary since all the server see's are HTTP requests anyway and can't really 'tell' if they are coming from a 'real browser' or not. Load tools also do a lot better job of instrumenting the system under test and can tell you if a disk queue is overloaded, a webserver is running out of memory, or any one out of hundreds of performance counters is out of a 'healthy' zone. In my experience it's that data that is key to figuring out where the constraints are so that the site can be improved to get to the performance levels required. You could use watir to measure the page load time etc when the server was under load, but in general a proper load tool, operating at the protocol level is far superior to trying to simulate a browser client. Even with a headless browser I think I'd have a hard time generating something like say oh 800 vusers worth of load from a single box, which is usually pretty easy with LR or VSTS. If you can see it 'going over the wire' in http format, you can generally simulate it with a good load tool. Usually it's mostly a matter of figuring out how to simulate any processing the client side code might be doing that actually affects what is sent over the wire. (most often it's just UI magic, and the key is knowing what parameters to look for and capture in the 'outbound' traffic from the server, so they can take their proper place in the requests sent back by the 'client' in subsequent requests. Yes you might be able to do some of that using watir-webdriver in a headless mode, but you would need to reinvent a lot of stuff such as how to feed unique parameters to each vuser, how to monitor transaction times and perf data, how to ramp up and ramp down load, etc etc On Jan 17, 9:03 am, Dan <dfra...@gmail.com> wrote: > So, maybe it's not practical with real browsers to do performance testing > with watir, but maybe it becomes more practical when you run the browser > headless with watir-webdriver? Where I am we use loadrunner, but we're > finding it challenging to deal with apps written using gwt. Obviously > there's some nice metrics and reporting that come out of loadrunner, but > for those cases where it's just not up to the technology like watir is, > maybe it's a feasible option. -- Before posting, please read http://watir.com/support. In short: search before you ask, be nice. watir-general@googlegroups.com http://groups.google.com/group/watir-general watir-general+unsubscr...@googlegroups.com
