Dimensional Data Modeling in the Modern Era
summary
Argues that classic dimensional modeling still pays off by example: an existing one big table SQL pipeline on DBT on Databricks (≈1,000 lines) gets rebuilt into a star schema, splitting into 21 models, adding 179 tests, expanding measures 1→6 and attributes 25→250, while analyst queries shrink to <20 lines. Result is comparable or better query time, far less redundancy, fewer maintenance requests and clear projected compute cost savings — a working, measurable migration and benefit.
AI-written summary · about
chapters (9)
- 0:00Introduction and agenda
- 1:25Data modeling overview
- 6:52Integration strategies
- 11:58Dimensional modeling basics
- 18:08Design and development steps
- 22:06Addressing common criticisms
- 25:30Real-world proof of value
- 30:24Key benefits and summary
- 32:26Closing thoughts
related talks
49mData Modelling Showdown: Unravelling the Mysteries of Different Data Modelling Approaches
39mPanel Discussion
43mScaling Data Stacks: How To Build a Great Data Model
16mGood Data Modeling & Effective Communication in Power BI
18mDemystifying Data Modeling
41mModern Data Architectures – What They Are and How to Implement?
24mMake data central feature
41mThe Unified Star Schema