1. Marcel   •  

    What does it tells uns? Do you think that a kimball based data model is is hard contrast to a mistake tolerance data model & response time? because there is only a small overlap of those 3.

    • Thomas Kejser   •  

      Hi Marcel

      I did not intend that the size of the overlap should be “to scale. My idea here was to show that it is indeed possible to create models that are tolerant to failure but which are NOT Kimball models. oFr example, a 3NF model is very forgiving of modeling mistakes, but will have horrible response time.

      The key here is that if you want all three, Kimball is an essential ingredient.

  2. Djamal   •  

    Yes, you’re right about DBUnit. I have used it some years ago at a pjcoert and we were able to load the test database based on data in MS Excel sheets. That was really cool!

Leave a Reply

Your email address will not be published. Required fields are marked *