Crippling Test Environments

Question

We have a web application that is sufficiently complicated that it's hard to accurately simulate production load.

One of our coping mechanisms has been to make sure that the hardware in the test environment is always slower that our production hardware, so load and performance problems will be more pronounced.

Are other folks out there doing this, what other strategies are you using?

I'm guessing that you're talking about *performance* testing? Or do you do all testing on slow hardware? — Jason Baker, Mar 24 '09 at 14:25
This doesn't sound like a great idea to me. Scaling a web app is all about removing bottlenecks, and by testing on different hardware than you run in production with, you may focus on the wrong bottleneck. — twk, Mar 24 '09 at 14:52

score 1 · Accepted Answer · answered Mar 24 '09 at 14:13

1

I have a 7-year-old 600mHz Celeron laptop with 256MB RAM that I keep around for performance profiling. For testing network latency, you could get a delay proxy (or write one trivially); it gets a packet, waits X milliseconds, then passes it on.

answered Mar 24 '09 at 14:13

DNS

37,249
18
95
132

score 1 · Answer 2 · answered Mar 24 '09 at 14:18

1

Fiddler can help you test low-bandwidth -- can simulate latency of slower networks.

answered Mar 24 '09 at 14:18

Lou Franco

87,846
14
132
192

score 0 · Answer 3 · answered Mar 24 '09 at 14:45

0

Hook up testmaker. Do some manual tests and let their proxy write the test script by recording what you did manually. Automate using the script(s). http://www.pushtotest.com/ Open source and free.

answered Mar 24 '09 at 14:45

Jay

13,803
4
42
69

Crippling Test Environments

3 Answers3