www.sum-it.nl > Courses
> DB Design course > Method > 3. Technical optimisation > Steps > Denormalise		Previous Previous Next Next
Nederlands · Search...

Denormalise

Denormalise from area to Reader See whether you can avoid an expensive join in a high frequency function by denormalising columns from an upper table to a lower one.

For example: Copy street name and city to reader.

Introduction 1. Scope 2. Logical Design
3. Technical Optimisation
Goal Optimisation steps Use ranges
Denormalise
Combine tables Store derived data Add indexes Sort 4. Implementation End
Attention Disadvantage of denormalisation: When the data changes in the one table, all copies in the many must be updated as well.

Example: When the street name changes all copies of the street in readers must be updated as well.

  • Focus denormalisation on data that hardly changes, such as street name. A street rarely changes name.
  • Focus on the high frequency functions that read a lot of data from various tables. (A list of subscriptions, including the address data for example).
  • Avoid fake optimisation. When you denormalise 9 data elements, but skip one essential one, the join will still be necessary.
Do not if:
  • it fails to increase performance
  • the software would get unacceptably complex.
Next: Combine tables