Optimisation Steps

  1. Use ranges
  2. Denormalise
  3. Combine tables
  4. Store derived data
  5. Add indexes
  6. Sort (clustered index)
For each of these actions you must follow the following rules:
  • Work bottom-up. Start optimisation at the core business, the tables for the high frequency functions. These are usually the tables at the bottom of the model.
  • Focus on one table at the time, but for all functions. Go to an upper table when you've done all tables below.
  • Always make a cost-benefit analysis. Each optimisation has it's price. Each optimisation must be beneficial in the end.