data-engineering interview questionsQuestion 1 of 6
Data Engineering interview question · Question 1 of 6
ETL vs ELT: what factors decide the choice?
Short answer
ETL transforms data before loading it into the target; ELT loads raw data first and transforms it inside the warehouse or lakehouse. I decide based on whether rules such as privacy require transformation before storage, where compute is cheapest and most scalable, whether we need to reprocess history from raw data, whether the formats suit SQL, and the team's skills. In practice many pipelines do light cleaning and masking on the way in and the main modelling after load.
Detailed explanation
| Factor | Favours ETL | Favours ELT |
|---|---|---|
| Compliance | Data must be masked or removed before it is stored | Raw storage is allowed under access control |
| Compute | Target cannot scale transformations | Warehouse/lakehouse scales SQL cheaply |
| Reprocessing | Rarely needed | Frequent rule changes, backfills |
| Data shape | Binary, deeply nested, non-SQL processing | Tabular or semi-structured data SQL handles well |
| Team | Strong software engineering | SQL-first analytics engineering |
Example answer structure
- Define both in one sentence each.
- Name two or three deciding factors for this scenario.
- Mention that hybrids are common, such as masking PII at ingestion, then ELT modelling.
Common mistakes
- Claiming one is always superior.
- Forgetting governance: “load everything raw” can violate data-protection rules.
Progress is saved in this browser only. No account needed.