Fabric Data Engineer Associate · 34% of the exam

Monitor and optimize an analytics solution: free practice questions

5 sample questions from our 60-question bank for this domain — answers and explanations included. These are the same scenario-based style as the real Microsoft exam.

1. A Fabric data engineer is troubleshooting an intermittent failure in a Fabric Data Pipeline. The pipeline has three sequential activities: a Lookup activity that reads a control table, a ForEach activity that iterates over the results, and inside the ForEach, a Copy Activity. The pipeline fails with error code '429 TooManyRequests' on the Copy Activity approximately 2 minutes into execution. The source is an external REST API. What is the most appropriate remediation?

A. Increase the Copy Activity timeout property from the default 12 hours to 24 hours to allow more time for the API to recover
B. Enable the 'Fault tolerance' setting on the Copy Activity to skip rows that cause errors and log them to a separate file
C. Configure the Copy Activity retry policy with a retry count of 3 and an exponential backoff interval, and reduce the ForEach batch count to limit concurrent API requests✓ Correct
D. Switch the pipeline trigger from a scheduled trigger to an event-based trigger to reduce the frequency of pipeline runs

Explanation

A 429 TooManyRequests error indicates the external REST API is rate-limiting the pipeline due to too many concurrent or too-frequent requests. Configuring an exponential backoff retry policy on the Copy Activity allows failed requests to retry after progressively longer waits, giving the API time to recover. Reducing the ForEach batch count lowers the number of concurrent Copy Activities hitting the API simultaneously, directly addressing the rate-limiting root cause. Option A is incorrect because increasing the timeout does not reduce the request rate or add retry logic; the API will continue to reject requests regardless of how long the pipeline waits. Option B is incorrect because fault tolerance skips erroring rows and writes them to a log—it suppresses the error rather than fixing the rate-limit issue, and would result in incomplete data loads. Option D is incorrect because changing the trigger frequency affects how often the pipeline runs in total, but within a single run, the ForEach still generates the same number of concurrent API calls.

2. A data engineer needs to configure a Fabric Data Activator alert to send a notification when the 5-minute rolling average of a 'CPU_Utilization' measure in a real-time KQL queryset exceeds 85%. The engineer opens Data Activator and creates a new Reflex. Which source should the engineer connect to in order to trigger alerts based on data already stored and queried from an Eventhouse?

A. Connect directly to the Eventhouse KQL database as a Data Activator source, using a KQL query to define the measure.✓ Correct
B. Export the KQL queryset results to a Lakehouse table first, then connect Data Activator to the Lakehouse table.
C. Connect Data Activator to an Eventstream that re-ingests the Eventhouse data back through Event Hubs.
D. Use a Fabric Data Pipeline with a Script activity to call the Data Activator REST API on a schedule.

Explanation

Data Activator (Reflex) supports a direct connection to Fabric Eventhouse KQL databases, allowing you to define a KQL query as the data source for trigger conditions. This is the most direct and appropriate method. Exporting to a Lakehouse adds unnecessary latency and complexity. Re-ingesting via Eventstream duplicates infrastructure and introduces additional lag. Using a pipeline with a REST API call is a workaround that requires custom development and does not use Data Activator's built-in source capabilities.

3. Northwind Traders ingests order data into a Fabric Eventhouse KQL database. The events table has more than 2 billion rows and is partitioned by ingestion time. A KQL analyst reports that queries filtering on 'region' (a string column with 12 distinct values) and a time range of the last 30 days consistently scan the entire table. The data engineer wants to improve query performance for this access pattern without rebuilding the entire ingestion pipeline. Which Eventhouse feature should the engineer configure?

A. Streaming ingestion policy — enables real-time row-level indexing on the 'region' column
B. Table update policy — transforms incoming data to pre-aggregate by region before storage
C. Partitioning policy — configures a hash partition on 'region' in addition to the existing time-based partition, directing queries to fewer shards✓ Correct
D. Row-level security policy — restricts data scanned per query based on the caller's region membership

Explanation

Configuring a partitioning policy with a hash partition on 'region' causes Eventhouse to co-locate rows with the same region value into the same extents (shards). Queries filtering on 'region' can then skip extents that do not match, dramatically reducing scan volume when combined with the time-range filter. Option A is incorrect because the streaming ingestion policy controls real-time vs. batched ingestion latency, not column-level indexing or partition pruning. Option B is incorrect because an update policy transforms data at ingestion time into a target table but does not add partition pruning to the source table. Option D is incorrect because row-level security controls which rows a user can see for access control purposes, not query performance or extent pruning.

4. A data engineer at Contoso notices that a Fabric Data Pipeline that loads data from Azure Data Lake Storage into a Lakehouse is intermittently failing. The pipeline activity error message reads: 'The remote server returned an error: (403) Forbidden.' The pipeline uses a Managed Identity for authentication. Which action should the engineer take FIRST to resolve this issue?

A. Regenerate the storage account access key and update the linked service connection string.
B. Verify that the Fabric workspace Managed Identity has been granted the Storage Blob Data Reader role on the Azure Data Lake Storage account.✓ Correct
C. Increase the pipeline timeout setting to allow more time for the authentication handshake to complete.
D. Switch the pipeline authentication method from Managed Identity to a service principal and provide new credentials.

Explanation

A 403 Forbidden error with Managed Identity authentication almost always indicates a missing or incorrect RBAC role assignment on the target resource. Granting 'Storage Blob Data Reader' (or 'Contributor') on the ADLS account to the Managed Identity resolves the authorization failure. — Option A is wrong because regenerating storage keys is only relevant when using key-based authentication, not Managed Identity. Option C is wrong because a 403 is an authorization error, not a timeout; increasing the timeout has no effect. Option D is wrong because switching authentication methods is a disruptive workaround rather than fixing the root cause, and the Managed Identity approach is correct—it just needs the right permission.

5. A data engineer is working with a Delta table in a Fabric Lakehouse. Over six months of incremental loads, query performance has degraded significantly. Running DESCRIBE DETAIL on the table shows 14,200 small Parquet files averaging 18 KB each. Which command should the engineer run to consolidate these files and improve read performance?

A. VACUUM delta.`/Tables/sales` RETAIN 0 HOURS
B. OPTIMIZE delta.`/Tables/sales`✓ Correct
C. ALTER TABLE sales COMPACT FILES
D. REORG TABLE sales APPLY (PURGE)

Explanation

OPTIMIZE is the Delta Lake command that compacts many small files into larger, optimally sized files (typically targeting ~1 GB each), directly addressing the small-file problem and improving query performance. — Option A (VACUUM) removes old, unreferenced data files based on a retention period; it does not compact files and running it with RETAIN 0 HOURS is dangerous as it can break time travel. Option C (ALTER TABLE … COMPACT FILES) is not a valid Delta Lake or Fabric SQL syntax. Option D (REORG TABLE … APPLY PURGE) is a SQL Server/Azure Synapse Dedicated Pool command that purges deleted rows in columnstore indexes; it does not apply to Delta tables in a Lakehouse.

55 more questions in this domain

Practice the full bank with instant grading, flashcards, and a timed mock exam.

Start practicing free