Today’s blog post starts with a question:
“In my continuous integration environment, running tests
currently takes 30 minutes. I
believe that a large percentage of time is spent in MySQL, and I want to know if putting MySQL
on tmpfs will cut that time down?”.
I have previously written about how to configure MySQL to be less durable in testing environments, but today I wanted to write about how we can really test the assumption that the question is based around. That is to say:
- We know that total time is 30 minutes.
- We assume MySQL is a large percentage of this.
- We know that this optimization will potentially allow us to avoid IO.
- We then want to know what amount of time MySQL spends waiting on IO during our test suite?
And the answer to …[Read more]