Alternatives for automated web application front-end performance testing [closed]

Answer 1

Since you have tagged ASP.Net, you might want to look at ANTS performance profiler- http://www.red-gate.com/products/dotnet-development/

Answer 2

It's worth having a look at Chrome Telemetry (http://www.chromium.org/developers/telemetry)

It's obviously Chrome only but it's what Adobe use to test the performance of TopCoat

Answer 3

To start with - i am using browsermob proxy for response times and codes, but indeed it does not give you anything like JS execution time. You already mentioned DynaTrace so I will not discuss this. What else is there - I have not used myself, but i have heard people using httpWatch. It integrates very vell with C# and is said to be capable of analyzing the performance of ajax calls.

Also, don't forget, that you can inject javascript with webdriver, so you can try to make use of boomerang or something like this.

Answer 4

http://html5testing.co uses phantom.js and casper.js together. They custom build a test that interacts with your app across any browser using webkit. You get reports with screenshots, can run the tests whenever you want, it does latency monitoring, and you can see all the previous tests you ran.

It's not free, but it's certainly a lot cheaper than all the time you could be putting into testing with a patchwork of stuff.

Answer 5

Have you looked at what your server will capture for you on your requests? Every web server has the ability to capture and record time in seconds (some in microseconds) for every single request made by your front end. You can pull the data on the backend and visualize it with any number of tools. Combine it with a free GEoIP database and you can even look at the data by country and city. Pay a little more for GEoIP data and you can look at mobile on mobile networks vs mobile on home or corporate networks to observe key differences in behavior by network connection (which you cannot control).

As to the obsession with front end performance, leave that to your functional testers. Optimization can be handled on a single PC with tools in the browser combined with something like Charles proxy to vary caching options and speeds. Once your application gets into the wild you are going to have very little control over the front end performance anyway. Rendering time will vary from PC to PC depending upon browser type and load on the individual PC. Have you ever seen what happens to a browser with just a few toolbars loaded and tracking all user actions? Even the fastest of sites with the most optimized pages slow to a crawl.

Do your job and optimize the page design, number of components, caching models, etc... and then for performance and scalability turn your attention to the server and away from the client.

Answer 6

My feeling on this subject is the best approach is to inject javascript into the page and have that report back metrics on the user experience. There are various approaches to this but they basically resolve down to either rolling your own code (more work but ultimately better results) or using a third party offering. New Relic have a pretty nice solution.

The problem with this approach is you still need something to exercise the browser in order for the javascript to get fired - it is hard to automate. That's the gotcha but other than that you have a reliable, consistent, maintenance-free method for recording and logging the user experience.