Prepare data
Free Microsoft Certified: Fabric Analytics Engineer Associate practice — 6 questions on Prepare data, with explanations. No sign-up. Full 12-question mixed test →
A Fabric lakehouse contains a fact table with 800 million rows that must be incrementally upserted every hour using complex merge logic based on multiple business keys and change-detection columns. The current process uses Dataflow Gen2 and is failing to complete within the required time window. Which approach BEST meets the requirement?
PySpark notebooks using Delta Lake MERGE scale efficiently for very large datasets with complex conditional upsert logic and can be scheduled/orchestrated via a pipeline; the Dataflow Gen2 mashup engine is not designed for this scale of transformation.
Your organization has parquet files stored in an existing Azure Data Lake Storage Gen2 account, managed and maintained by another team outside Fabric. You need these files to be queryable as native tables from within a Fabric lakehouse without copying or duplicating the underlying data. What should you do?
OneLake shortcuts create a virtualized reference to external storage such as ADLS Gen2, exposing the files as native lakehouse tables without duplicating or moving the underlying data.
A Fabric Warehouse contains a large fact table. Reporting queries that join this fact table to a dimension table on a non-key column are running slowly. The clustered columnstore index and row-group elimination are functioning correctly, but the query optimizer keeps choosing an inefficient join plan. What should you do to improve performance?
Fabric Warehouse auto-creates some statistics, but columns used in joins or filters that lack up-to-date statistics can cause the optimizer to pick a poor plan; manually creating statistics with CREATE STATISTICS gives the optimizer better cardinality estimates.
Your company ingests high-velocity IoT sensor telemetry (50,000 events/second) and requires sub-second analytical queries using time-series functions such as series_decompose() over the streaming data, with results feeding real-time dashboards. Which Fabric item should you use to store and query this data?
An Eventhouse (KQL Database) within Real-Time Intelligence is purpose-built for high-velocity streaming ingestion via Eventstream and supports native KQL time-series functions like series_decompose() with low-latency query performance.
You are modeling a semantic model using a star schema. The Products dimension and the Promotions dimension have a many-to-many relationship: a product can be part of multiple promotions, and a promotion can apply to multiple products. Which modeling approach BEST resolves this while maintaining accurate filtering behavior?
A bridge (associative) table holding the valid key combinations is the standard star-schema pattern for resolving many-to-many relationships between two dimensions while preserving correct, unambiguous filter propagation.
A data engineer uses Copilot in a Fabric notebook to generate PySpark code that aggregates total sales by region from a sales fact table. The code runs successfully and produces output, but the business requirement specifies that returned orders (negative quantity) must be excluded from the total. What should the engineer do?
Copilot-generated code must always be reviewed and validated against actual business requirements; Copilot cannot infer unstated business rules such as excluding returns, so verification and correction before deployment is required.
Ready for the real thing?
The full course: two full-length practice tests, video lessons for every exam domain, hands-on labs and detailed explanations.
$109.99 $34.99 with code FREETEST33 — valid through September 16.
Get my $34.99 deal →