← Interview & exam prep
Technical interview questions
Microsoft Fabric interview questions
15 questions from beginner to advanced, each with the correct answer and a one-line explanation of why. Study them here, or take them as a timed test.
15 questionsBeginner 8 · Intermediate 7Answers + explanations
Beginner 8 questions
1OneLake is…
- A single storage account per report
- A capacity SKU
- A visual
- A gateway
- One unified, tenant-wide logical data lake
Show answerHide answer
E · One unified, tenant-wide logical data lake OneLake is the single, organization-wide data lake underpinning Fabric (OneDrive for data).
2Direct Lake storage mode…
- Reads Delta/Parquet from OneLake without import or DirectQuery translation
- Requires a gateway
- Caches in the browser
- Imports data on a schedule
- Only works on-premises
Show answerHide answer
A · Reads Delta/Parquet from OneLake without import or DirectQuery translation Direct Lake reads columnar files directly, giving import-like speed without a refresh copy.
3A OneLake 'shortcut'…
- Is a keyboard macro
- Copies data nightly
- References data in place without copying it
- Compresses tables
- Deletes duplicates
Show answerHide answer
C · References data in place without copying it Shortcuts virtualize external/other data into OneLake without duplication.
4A Lakehouse combines…
- Lake storage with a SQL analytics endpoint and table semantics
- Only files
- Only a database
- Two gateways
- Two warehouses
Show answerHide answer
A · Lake storage with a SQL analytics endpoint and table semantics A Lakehouse gives you files + Delta tables + a read-only SQL endpoint in one item.
5The Delta Lake format is essentially…
- JSON only
- Parquet plus a transaction log enabling ACID
- A zip archive
- An Excel file
- Plain CSV
Show answerHide answer
B · Parquet plus a transaction log enabling ACID Delta = Parquet data files + a JSON transaction log for ACID and time travel.
6Fabric compute is billed in…
- Reports
- Capacity Units (CU) via F SKUs
- Users
- Refreshes
- Gigabytes
Show answerHide answer
B · Capacity Units (CU) via F SKUs Fabric capacities (F2, F64, …) are measured in Capacity Units.
7A Fabric Notebook primarily runs on…
- SSAS
- VertiPaq
- Excel
- Apache Spark
- A gateway
Show answerHide answer
D · Apache Spark Notebooks execute Spark (PySpark/Scala/SQL) against Lakehouse data.
8Data Factory in Fabric provides…
- A semantic model
- Only visuals
- RLS
- Pipelines and dataflows for orchestration and ETL
- A gateway
Show answerHide answer
D · Pipelines and dataflows for orchestration and ETL Fabric Data Factory offers pipelines and Dataflows Gen2 for moving and transforming data.
Intermediate 7 questions
9Direct Lake typically FALLS BACK to DirectQuery when…
- Unsupported features or data exceed the mode's guardrails
- A slicer is used
- The user is offline
- RLS is applied
- The report is large
Show answerHide answer
A · Unsupported features or data exceed the mode's guardrails Exceeding guardrails (size/features) triggers automatic DirectQuery fallback.
10In Fabric, a Warehouse (vs a Lakehouse) offers…
- Only Python
- No SQL
- Only Spark
- Files only
- Full T-SQL with multi-table read/write transactions
Show answerHide answer
E · Full T-SQL with multi-table read/write transactions The Warehouse is a full T-SQL engine with read/write; the Lakehouse SQL endpoint is read-only.
11The medallion architecture layer order is…
- Raw → Gold → Bronze
- Silver → Bronze → Gold
- Bronze → Silver → Gold
- Bronze → Gold
- Gold → Silver → Bronze
Show answerHide answer
C · Bronze → Silver → Gold Bronze (raw) → Silver (cleaned/conformed) → Gold (curated/serving).
12V-Order optimization…
- Reorders Parquet for faster VertiPaq/Direct Lake reads
- Deletes duplicates
- Adds indexes to SQL
- Encrypts files
- Sorts rows alphabetically
Show answerHide answer
A · Reorders Parquet for faster VertiPaq/Direct Lake reads V-Order is a write-time Parquet optimization that speeds columnar reads.
13Dataflows Gen2 can write their output to…
- Only the Power BI model
- Nowhere; they are read-only
- Lakehouse/Warehouse (and other) destinations
- Only CSV on disk
- Only Excel
Show answerHide answer
C · Lakehouse/Warehouse (and other) destinations Gen2 dataflows support configurable output destinations like Lakehouse and Warehouse.
14The SQL analytics endpoint of a Lakehouse is…
- A Spark cluster
- Read/write T-SQL
- Read-only T-SQL over the Delta tables
- A gateway
- A dataflow
Show answerHide answer
C · Read-only T-SQL over the Delta tables You can query (SELECT) the Lakehouse tables via T-SQL, but writes go through Spark/Delta, not this endpoint.
15A default semantic model over a Lakehouse uses which storage mode?
- Import
- Dual
- DirectQuery
- Direct Lake
- Composite
Show answerHide answer
D · Direct Lake Fabric auto-creates a Direct Lake semantic model over the Lakehouse tables.
What changed lately? Latest Microsoft Fabric news, releases & questions →
More technical interview questions