Data Warehouse Toolkit Chapter VIII - notes

• Customer contact table:
  • Split it down to as many elemental parts as possible.
• Use unicode as encoding - otherwise handling multiple encodings will be a nightmare
• If warehouse is to be used internationally - translate the reports, and not all content of warehouse
• According to authors, data mining teams should get exports from warehouse, so they can do computations on their own servers. (imagine exporting couple hundred TB :))
• Solution for tracking user’s activity - create a “steps” dimension table which will refer to what user can do. Then link each event to that dimension.