Skip to main content

21.4 Data Layers & Modeling Zones

A layered architecture prevents raw acquisition logic, business semantics and consumption-specific calculations from becoming one tangled model.

1. Acquisition​

Purpose: land or access source-aligned data.

2. Harmonization​

Purpose: standardize keys, types, reference data and integration logic.

3. Semantic​

Purpose: define reusable facts, dimensions, measures and business meaning.

4. Consumption​

Purpose: expose focused analytical models and consumer-specific contracts.

5. Product​

Purpose: package governed, discoverable reusable data for broader consumption.

These are logical responsibilities, not mandatory physical layers or separate spaces.

Grain discipline​

Every fact-like entity must state its grain.

Example: One row = one purchase-order item.

Never mix grains casually inside a model.

Anti-pattern​

Raw + transformation + semantic + dashboard logic inside one giant view.

It becomes difficult to test, reuse, optimize and govern.

Architect rule​

Layer when responsibility changes, not merely because a diagram has room for another box.