An organisation needs to bring data from a transactional database, a SaaS application and an existing data warehouse into its lakehouse. What is the correct approach?
Select an answer to reveal the explanation.
Short Explanation
Use the connector that fits each source, then decide per source whether to ingest or federate based on latency and use case. Forcing one connector on everything gives awkward sources a worse integration than they need.
Full Explanation
The correct approach is to use the supported connector appropriate to each source, choosing ingestion or federation per source according to its latency and use-case requirements. Sources differ in protocol, freshness expectation and volume, so the platform provides a range of connectors and both movement and federation patterns. The design decision is made per source: how fresh must this be, how large is it, and does the use case justify copying it.
The approach of standardising on a single connector type and adapting every source to it is incorrect because one connector cannot address the protocols and semantics of unrelated systems, and forcing it produces brittle workarounds.
The approach of having analysts paste representative source data into query worksheets is incorrect because manual pasting does not scale, is not repeatable, and produces data with no provenance.
The approach of requiring each source owner to export screenshots of their data for upload is incorrect because screenshots are images, not data - they cannot be queried, joined or governed.
For the exam, the integration answer is per-source connector choice driven by latency and use case.