Skip to main content
Software Engineering

You are not logged in. Your edit will be placed in a queue until it is peer reviewed.

We welcome edits that make the post easier to understand and more valuable for readers. Because community members review edits, please try to make the post substantially better than how you found it, for example, by fixing grammar or adding additional resources and hyperlinks.

Required fields*

8
  • 7
    Hi Yosi, I understand your point. Normalization is foundational in really understanding the theory of relational databases and has real application in practice, so it isn't surprising that it is a big topic in courses. Good engineers should understand it and understand when it should be applied. The thing that doesn't seem to be covered in the course work is that selectively denormalizing can yield a lot of benefit and some problems really don't lend themselves to normalized models. Commented Sep 29, 2013 at 17:52
  • 1
    What about data consistency ? For example if you have the shop name in every sales' detail, then you can potentially have different contradicting descriptions, whereas if data is normalized, the shop name appears only one (in the shop table) and there's no place for inconsistency. Commented Sep 30, 2013 at 15:55
  • 1
    I agree. I think normalization gets over used at times by DBAs that have been taught that this is the best design. I've always suggested that the DBAs can normalize the tables in the ETL all they want, but when it comes to the tables the UI references, I need tables that are easy to query without excessive joins. I've run into tables that were so over-normalized, so could barely troubleshoot user issues without spending HOURs troubleshooting. Commented Oct 7, 2013 at 19:37
  • 1
    Au contraire, analytics is insanely difficult if you're unable to start from a normalized model. I just had to go through this exercise, and it was hell. Application developers should never assume that a denormalized schema is going to be suitable for analytics needs. And as for point #3 against normalization, it's a problem that's almost trivially solved by materialized/indexed views. Commented Oct 26, 2013 at 21:52
  • 1
    And #2 sounds reasonable but strains credulity in practice - I cannot remember seeing a single instance in my 10+ years where constraints were actually thoroughly enforced by the application. More often, developers either incorrectly equate business rules to data integrity or use the fact that ORMs theoretically can enforce relational constraints as an excuse not to do it anywhere at all. Maybe I'm just being cynical, but all of my career experience has taught me that statements like "the application will enforce data integrity" are enormous red flags. Commented Oct 26, 2013 at 21:54