RDB/Site Stress Testing
paddy last edited by
So, when you build a new design, or try something fairly new, what is your preferred method of stress testing?
I have some good benchmarking code and such that I like to use to find
bottlenecks, but thats not the same as throwing the equivelent of
peak-use traffic at a database after its been filled with relevant data.
Sadly, most of the time, I'll come into a project that is already up
and going with a lot of data and get paid to "fix" the really slow
code, find a ratsnest in the overpriced PHP/MySQL app they bought, and
when I get a page from 13s to .8s on a wide search I am happy, and I
know its already handling stress okay, but thats not the same as
benchmarking during prototype development.
What solutions do people usually implement to solve these sorts of
problems? How do you calculate load vs. performance in advance of
getting tons of traffic and a heavy load?