ETL, ELT, and Incremental Loading: 10 practice questions

Question 1 of 10

Where does transformation occur in ETL?

1. ADuring extraction, without a load step
2. BAfter loading, only inside the analytical target with raw data already stored
3. CBefore loading, on processing outside the analytical target
4. DAfter serving, inside the reporting client

Question 2 of 10

What is the defining order of ELT?

1. ALoad, extract, then transform at the source
2. BExtract, transform, then load the result
3. CTransform, serve, then extract the report
4. DExtract, load, then transform in the target

Question 3 of 10

What does a watermark support in an incremental load?

1. ASelecting records later than the last processed value
2. BTreating every source record as permanently new
3. CRebuilding every target row from the full source without any saved progress
4. DTransforming raw data only after report delivery

Question 4 of 10

Which constraint most directly favors ETL?

1. AAnalysts need to repeatedly transform retained raw inputs
2. BThe target can scale transformations in parallel
3. CMany input formats should land before their use is known
4. DSensitive raw fields must be masked before entering the target

Question 5 of 10

Why can ELT simplify infrastructure?

1. ANo governance is needed after raw data is loaded
2. BExtraction and loading are removed from the pipeline
3. CTransformation runs in the target instead of a separate processing server
4. DThe target never stores untransformed data

Question 6 of 10

Which loading pattern adds new records but does not revise existing target rows?

1. AFull refresh loading
2. BAppend-only loading
3. CChange-based upsert loading
4. DComplete target replacement

Question 7 of 10

What distinguishes a change-based incremental load from a full refresh?

1. AIt can insert records but cannot detect modifications
2. BIt processes records new or modified since saved progress
3. CIt refuses to retain any state between runs
4. DIt reloads the complete source on every run

Question 8 of 10

A regulated target must never receive unmasked personal identifiers, its schema is fixed, and a separate processing tier already exists. Analysts do not need the raw values. Which pattern fits best?

1. ALoad raw identifiers first and mask them later in the target
2. BSkip transformation and expose source records directly
3. CRetain all raw identifiers for repeated target-side exploration
4. DTransform and mask before loading the fixed target

Question 9 of 10

A cloud warehouse can scale compute, incoming data has varied formats, and analysts need to derive new models repeatedly without re-extracting sources. Raw access is governed inside the target. Which pattern is strongest?

1. ADiscard raw inputs after producing one predetermined model
2. BLoad the raw inputs and transform them in the target as needed
3. CTransform every use case on a fixed external server before loading
4. DRun a full source extraction for every analytical question

Question 10 of 10

A source contains 500 million rows, about 0.2 percent change daily, updates to old keys matter, and the pipeline can store a reliable last-modified watermark. Which load design minimizes repeated work without losing revisions?

1. AReplace the entire target from a complete source extract daily
2. BAppend every extracted row as a new target record
3. CSelect and apply rows beyond the saved change watermark
4. DLoad only newly created keys and ignore older modified keys