Skip to content
DP-700 Fabric Engineer Hub

Exam DP-700: Scenario-Based Practice Questions

Exam DP-700: Scenario-Based Practice Questions

[!NOTE] These practice questions are model-authored study aids, designed to replicate the scenario complexity and technical difficulty of official Microsoft DP-700 certification items.

Question 1 (Domain 1: Storage & Architecture)

Scenario: An enterprise data platform team needs to ingest real-time sales transactions from an on-premises Oracle database into Microsoft Fabric. The data must be accessible for near-real-time Power BI reporting using Direct Lake mode with minimal compute latency. Which architecture should you implement?

  • A) Ingest data via Dataflows Gen2 writing to a Lakehouse Files/ folder.
  • B) Ingest data using Data Factory Copy activity writing as Delta tables into Lakehouse Tables/ with V-Order enabled.
  • C) Create a Fabric Warehouse and load data using T-SQL INSERT INTO statements every hour.
  • D) Ingest data into an Eventstream writing to an ADLS Gen2 external shortcut as CSV.

Correct Answer: B
Rationale: Direct Lake mode in Power BI requires tables to be stored in Delta Lake format under the managed Tables/ directory of a Lakehouse or Warehouse with V-Order optimization enabled. Writing to Files/ or raw CSV does not expose Delta metadata to Direct Lake.

Question 2 (Domain 2: PySpark & Delta Optimization)

Scenario: You are maintaining a large Delta table (fact_telemetry) in a Fabric Lakehouse that receives hourly append batches. Over several months, query performance against the table has degraded significantly due to hundreds of small 5MB parquet files. In addition, users need to perform time travel queries going back 14 days. Which two commands should you execute to restore optimal performance while preserving the business requirements? (Select two)

  • A) OPTIMIZE fact_telemetry VORDER;
  • B) VACUUM fact_telemetry RETAIN 0 HOURS;
  • C) VACUUM fact_telemetry RETAIN 336 HOURS;
  • D) ALTER TABLE fact_telemetry SET TBLPROPERTIES ('delta.deletedFileRetentionDuration' = 'interval 1 days');
  • E) DROP TABLE fact_telemetry;

Correct Answer: A and C
Rationale: OPTIMIZE ... VORDER compacts small parquet files into optimal ~1GB files and organizes columnar storage for fast queries. VACUUM ... RETAIN 336 HOURS (14 days $\times$ 24 hours = 336 hours) cleans up obsolete physical files older than 14 days while safely preserving the 14-day time travel retention SLA. Vacuuming with 0 hours would permanently delete all historical files and break time travel.

Question 3 (Domain 2: Real-Time Intelligence & KQL)

Scenario: You need to process a continuous IoT sensor stream arriving in a Microsoft Fabric KQL Database. The raw JSON telemetry payload contains invalid sensor readings where Temperature < -50 or Temperature > 150. You want to automatically sanitize and store clean records into a curated target table at the moment of ingestion without running periodic batch jobs. What should you configure?

  • A) A scheduled Data Factory pipeline running every 5 minutes.
  • B) A KQL Update Policy attached to the curated table referencing a custom KQL filtering function.
  • C) A Power BI scheduled incremental refresh policy.
  • D) A Fabric Notebook scheduled via Apache Airflow.

Correct Answer: B
Rationale: A KQL Update Policy runs automatically and transactionally at ingestion time as new data lands in the source staging table, invoking a transformation function to populate the destination curated table with sub-second latency.

Question 4 (Domain 1: Security & Governance)

Scenario: A company requires that data analysts in the Finance_Analytics group can query aggregated sales metrics from a Fabric Lakehouse using T-SQL in the SQL analytics endpoint. However, analysts must NOT be allowed to read raw unmasked parquet files directly through Spark or Azure Storage APIs. How should you configure permissions?

  • A) Grant Admin role in the workspace.
  • B) Grant ReadAll item-level permission on the Lakehouse.
  • C) Grant Read permission on the Lakehouse and configure SQL Row/Column-Level Security on the SQL analytics endpoint.
  • D) Share the underlying ADLS Gen2 storage account key with the group.

Correct Answer: C
Rationale: Read permission allows users to query data through the SQL analytics endpoint where SQL security rules (RLS, CLS, masking) are enforced. Granting ReadAll gives direct storage access to raw OneLake Delta files, bypassing SQL endpoint security mechanisms.

Question 5 (Domain 3: Capacity & Monitoring)

Scenario: Your Fabric tenant administrator reports that interactive Power BI reports are receiving HTTP 429 errors (“Too Many Requests”) during peak morning hours. Background ETL pipelines are scheduled to run continuously during this same window. What capacity mechanism is causing the interactive queries to fail?

  • A) Interactive operations exceeded their 5-minute smoothing window and reached the 60-minute cumulative over-consumption rejection threshold.
  • B) Background batch operations exceeded their 24-hour window and were rejected first.
  • C) The OneLake storage limit of 100 TB was exceeded.
  • D) Git integration failed to sync workspace commits.

Correct Answer: A
Rationale: In Fabric Capacity throttling, interactive operations are smoothed over a 5-minute window and are throttled earlier than background operations (first delayed after 10 min debt, then rejected after 60 min cumulative over-consumption). Background batch jobs are smoothed over a 24-hour window.