DP-700 sample exam questions

A data pipeline Copy activity can use an on-premises data gateway connection to read from SQL Server and write to a Fabric warehouse. Dataflow Gen1 cannot target a Fabric warehouse, a KQL queryset only queries KQL databases, and notebooks cannot use the on-premises data gateway.

Workspace roles grant access to every item, so isolate the PII: create Workspace2 with Lakehouse2, migrate the Employee table there, and give the engineer Contributor on Workspace1 so they can write to Customer. Viewer cannot write, and roles on Workspace2 would expose the Employee data.

Fabric Monitor (the Monitoring hub) lets you open the notebook's Spark application run details, where the Spark and Delta runtime versions used for the execution are shown. Real-Time hub and OneLake data hub list data sources, and the Admin monitoring workspace and Capacity Metrics app track usage and capacity, not run-level Delta versions.

notebook.runMultiple accepts a DAG whose activities array lists each notebook, and a dependencies property makes Load_Orders wait for Load_Salesperson. Note that mssparkutils is now the legacy name; Microsoft recommends notebookutils, which offers the same method.

Only the changed items (Semanticmodel1 26 MB, Semanticmodel2 14 MB and Report2 12 MB, 52 MB total) are committed, and their combined size exceeds the per-commit size limit Fabric enforces for GitHub repositories. Splitting them into two selective commits keeps each under the limit, so two commits is the minimum.

A scheduled job gains nothing from high concurrency, and starter pools need no sizing and bill only while a session runs. Multi-node custom pools give several analysts enough capacity at once. This item is debatable, because Fabric offers no separate serverless Spark environment option to compare.

The eventstream Group by operator computes aggregations over a time window and supports functions such as average, min, max, sum, count, percentile and standard deviation. The Aggregate operator only supports sum, minimum, maximum and average, while Expand and Union do not compute statistics.

In a Delta merge, whenMatchedUpdate changes rows found in both tables, whenNotMatchedInsert adds rows found only in the source, and whenNotMatchedBySourceUpdate changes rows found only in the target. The exhibit's condition uses a greater-than comparison, which would actually need less-than for older rows.

A Lookup activity runs a query against the control table and returns its rows, whereas Get Metadata only returns item properties. ForEach loops over a known list, whereas Until repeats on a condition. The inner Copy data activity uses each row's schema and table values.

A data pipeline Copy activity provides the highest throughput for large file ingestion without transformations and supports storage event triggers that start the run when a new file arrives. Eventstreams process event data, environments store Spark settings, and KQL querysets only run queries.

Eventstreams match Stream Analytics and Event Hubs experience, since they move events from sources to destinations with no code. Dataflow Gen2 offers Power BI users the familiar low-code Power Query experience. Notebooks are the only listed item that runs R code, through SparkR or sparklyr.

Writing with format delta and saveAsTable("USCustomers") registers a managed Delta table in the lakehouse in one statement, and overwrite replaces the content with the filtered US customers. Parquet format does not create a Delta table in the Tables area, and append would duplicate rows on reruns.

2,000 files of 1 MB is the small-file problem: OPTIMIZE compacts them into larger files (with V-Order applied), and VACUUM removes the old unreferenced files. Disabling V-Order would make reads slower, not faster.

In the notebook's Spark application run details, the Run series view shows a timeline chart of the application's runs and their jobs, which suits a notebook that runs repeatedly. Monitoring hub lists runs without a job timeline, Real-Time hub is for streaming data, and the job history list is not a chart.

The community answer C is wrong because Log Analytics integration is the older Power BI feature for semantic model logs. Fabric workspace monitoring is turned on from Workspace settings by adding a monitoring eventhouse, which automatically collects logs and metrics for workspace items into a KQL database. ADLS Gen2 or a custom eventstream require more effort.