1. 25
  1.  

  2. 4

    Don’t care about fault, just make it better.

    1. 2

      Agreed, there is rarely only a single root cause.

      1. 2

        Three main things identified as improvements:

        1. upgrade your JDK
        2. premature optimization is the root of all evil
        3. RTFM
        4. automated performance monitoring in the testing env.

        Note: I’m the author of the linked article.

        1. 2

          premature optimization is the root of all evil

          I do not think this can be the takeaway.

          There is no way to not prematurely optimize if you are trying for anything. You read the documents and choose the options that seem nicest.

          1. 2

            I don’t agree that was a case of premature optimisation. You had a parameter you were choosing the value for. You considered the options and chose the one that made sense.

            I think premature optimisation would be if you spent time analysing just that bit and improving it (how about a kernel timer that causes thread activation to do the work?) without any evidence that this area needs attention/improvement.