Study Databricks DE-PRO Dimensional Modeling: key concepts, common traps, and exam decision cues.
The modeling questions at this level are rarely abstract. They ask whether the table shape and layer purpose match the downstream consumer and operational boundary.
| Layer or pattern | Better first reading |
|---|---|
| bronze | raw landing and replay-friendly fidelity |
| silver | cleaned, conformed, and reusable analytical foundation |
| gold | serving-ready outputs for business consumption |
| dimensional modeling | shape data for understandable downstream analytics and reporting |
| If the downstream need is… | Stronger first answer |
|---|---|
| preserve raw fidelity and replay options | bronze |
| create reusable cleaned foundations | silver |
| serve BI or business-ready outputs | gold |
| make downstream reporting easier to reason about | dimensional-model thinking |
Medallion names matter because they reflect different operational purposes, not because they are fashionable labels.
| If the stem says… | Strong reading |
|---|---|
| “business intelligence and reporting” | think gold and serving fit |
| “support downstream analytics cleanly” | silver or dimensional-model design may matter |
| “maximize collaboration across teams” | table organization and layer purpose affect usability and governance |
Using the wrong layer for the job creates recurring problems:
DE-PRO usually rewards the answer that keeps each layer honest.
| Trap | Better rule |
|---|---|
| using bronze as a general-purpose serving layer | bronze is for raw fidelity, not polished consumption |
| pushing all privacy and quality logic to the last layer | controls should appear at the right boundary earlier |
| treating medallion layers as only naming conventions | each layer has an operational purpose |
| Scenario clue | Stronger answer shape |
|---|---|
| “raw fidelity and replay” | bronze |
| “cleaned reusable analytics base” | silver |
| “BI-ready serving output” | gold |
| “shape tables for understandable reporting” | dimensional-model thinking |
Serving-layer modeling questions usually test whether you can distinguish raw fidelity from business-ready fit. Bronze preserves raw ingest. Gold is usually the serving-oriented layer for BI and curated outputs. If the requirement is easier downstream analytics, think table fit and dimensional-model instincts, not just “the pipeline landed successfully.” DE-PRO rewards downstream usability.