Spark engineers need files and tables in one place. Which store fits?
Why D is the answer
A lakehouse
A lakehouse combines files and Delta tables for Spark workloads.
In detail
- A lakehouse holds files and tables, and works with Spark.
- Lakehouse tables also get a SQL analytics endpoint for reading.
Use case
An illustrative scenario, not a real incident.
Engineers ingest raw JSON files and curate Delta tables from Spark notebooks, keeping both in one lakehouse.
See it in different ways
flowchart TD
A{"Main workload"} -->|"Spark and files"| B["Lakehouse"]
A -->|"T-SQL and transactions on tables"| C["Warehouse"]
A -->|"Streaming and time-series with KQL"| D["Eventhouse"]mindmap
root((Data stores))
Lakehouse
Spark
Files and tables
Warehouse
T-SQL
Transactions
Eventhouse
KQL
Streaming telemetrymindmap
root(("Lakehouse"))
n1["Files"]
n2["Raw JSON"]
n3["CSV"]
n4["Tables"]
n5["Delta tables"]Key point: A lakehouse holds files and tables.
Why the other options are wrong
- A. A KQL database
- A KQL database is for telemetry.
- B. A semantic model
- A semantic model serves reports.
- C. A warehouse
- A warehouse is T-SQL centered.