Warehouse, Lake, and Lakehouse: 10 practice questions

Question 1 of 10

Which workload is the clearest fit for a data warehouse?

1. AGoverned SQL reporting over integrated historical tables
2. BLanding raw source files without a defined schema
3. CCapturing individual application transactions for immediate updates
4. DStoring raw sensor media for several future processing engines

Question 2 of 10

What is the main responsibility of a serving layer?

1. ARename all storage layers to one mandatory industry vocabulary
2. BCapture operational transactions before analytical ingestion
3. CPublish clean, enriched layouts designed for downstream consumers
4. DLand every source in its original form without quality checks

Question 3 of 10

What does a lakehouse aim to combine?

1. ARaw file retention with no schema or governance capabilities
2. BScalable lake storage with warehouse-style management and serving
3. COperational transaction capture with no analytical workloads
4. DA reporting client with a dedicated source database for each user

Question 4 of 10

Why is cataloging essential in a data lake?

1. AVaried raw data must be discoverable and understandable to remain usable
2. BCataloging removes access controls from raw data
3. CCataloging forces every analytical engine to use one physical compute cluster
4. DCataloging converts every unstructured object into a warehouse table automatically

Question 5 of 10

Which sequence correctly describes progressive data refinement?

1. APublish raw inputs, remove consumer layouts, then restore source transactions
2. BCreate serving tables, bypass ingestion, then apply source-side transaction locks
3. CLand raw inputs, validate and integrate them, then publish consumer-ready data
4. DAggregate reports, discard validation, then land untracked source files

Question 6 of 10

How should a team interpret bronze, silver, and gold layer names?

1. AAs file formats that replace Parquet and ORC at each quality level
2. BAs common local labels whose actual guarantees still require documentation
3. CAs a universal standard with identical guarantees on every platform
4. DAs mandatory compute services for ingestion, transformation, and reporting

Question 7 of 10

Why can an excessive number of small files hurt analytical reads?

1. ASmall files eliminate all opportunities to process data in parallel
2. BFile listing, opening, and task scheduling add overhead before useful scanning
3. CSmall files automatically remove schemas from warehouse tables
4. DSmall files force every columnar format to become row-oriented

Question 8 of 10

A company must retain raw images and logs, run machine learning and SQL, and later publish governed tables for BI on shared storage. Which architecture best matches the combined workload?

1. AAn unmanaged data lake with no catalog, access controls, or validated serving data
2. BAn operational database that serves transactions, raw media, and all historical analytics directly
3. CA warehouse that rejects non-tabular data and exposes only predeclared BI tables
4. DA lakehouse with raw landing, managed refinement, and governed serving layers

Question 9 of 10

Queries usually filter by event date, rarely by event ID, and daily volume is high enough for parallel scans. Which partition strategy is strongest?

1. AUse date boundaries and size files within each date for useful parallel work
2. BUse event ID boundaries because high cardinality guarantees less listing overhead
3. CCreate one partition per event ID and keep every resulting file as small as possible
4. DKeep one file for all dates because common date filters cannot benefit from layout

Question 10 of 10

A raw layer receives files exactly as delivered, a validated layer enforces quality, and a serving layer publishes aggregates. Another platform uses different layer names. What should the teams standardize?

1. AOne physical file format, while leaving quality and access expectations undefined
2. BThe responsibilities and guarantees of each state, while allowing local names
3. COne compute engine, while removing the distinction between raw and serving data
4. DThe words bronze, silver, and gold, while allowing each word to mean a different state