Home › Key concepts › Normalised vs denormalised
Key concept · Walk (intermediate) · the same on every platform
Normalised vs denormalised
Store each fact once, or repeat it to make reading faster.
What it is
Normalised data (as in app databases) stores each fact once and links tables by keys, which keeps updates correct. Denormalised data repeats values, such as the product name on every sale row, to avoid joins and speed up reading.
Why it matters
Apps write a lot and want normalised data. Analytics reads a lot and usually wants some denormalising.
An example
The app keeps products in one table; the gold sales table copies the product name and category onto each row.
The common mistake
Copying the app's normalised design straight into analytics and making every report join twelve tables.
Where you meet it on each platform
The idea stays the same; only the product name changes.
| Job | AWS | Azure | Google Cloud | Databricks | Open source |
|---|---|---|---|---|---|
| App databasewhere an app keeps its own rows | RDS Postgres / DynamoDB | Azure Database for PostgreSQL / Cosmos DB | Cloud SQL / Firestore | Lakebase (Postgres) | Postgres |
| Warehousetables built for analysis, fast and governed | Amazon Redshift | Fabric Warehouse | BigQuery | Databricks SQL warehouse | DuckDB or Postgres |
Related ideas
Where it shows up
Part of the Walk level on these paths: Data Engineering (shaping data).