Databricks DATABRICKS-CERTIFIED-DATA-ENGINEER-ASSOCIATE Online Practice Questions and Exam Preparation

Databricks DATABRICKS-CERTIFIED-DATA-ENGINEER-ASSOCIATE Online Questions & Answers

• Question 91:

  A data engineer is processing ingested streaming tables and needs to filter out NULL values in the column from the raw streaming table orders_raw and store the results in a new table order_datetime using DLT.orders_valid Which code snippet should the data engineer use?

  

  A. Option A
  B. Option B
  C. Option C
  D. Option D
  A. Option A

  ---

  Explanation

  The correct approach is to use a DLT expectation with constraint CONSTRAINT valid_date EXPECT (order_datetime IS NOT NULL) ON VIOLATION DROP ROW This ensures rows with NULL order_datetime are filtered out automatically during streaming ingestion while keeping the pipeline valid.

• Question 92:

  A Databricks workflow fails at the last stage due to an error in a notebook. This workflow runs daily. The data engineer fixes the mistake and wants to rerun the pipeline. This workflow is very costly and time- intensive to run. Which action should the data engineer do in order to minimise downtime and cost?

  A. Re-run the entire workflow
  B. Repair run
  C. Restart the cluster
  D. Switch to another cluster
  B. Repair run

  ---

  Explanation

  Repair run in Databricks allows rerunning only the failed tasks in a workflow instead of re-running the entire pipeline, minimizing downtime and reducing costs.

• Question 93:

  A data engineer is running code in a Databricks Repo that is cloned from a central Git repository. A colleague of the data engineer informs them that changes have been made and synced to the central Git repository. The data engineer now needs to sync their Databricks Repo to get the changes from the central Git repository.

  Which of the following Git operations does the data engineer need to run to accomplish this task?

  A. Merge
  B. Push
  C. Pull
  D. Commit
  E. Clone
  C. Pull

  ---

  Explanation

  To sync a Databricks Repo with the changes from a central Git repository, the data engineer needs to run the Git pull operation. This operation fetches the latest updates from the remote repository and merges them with the local repository.

  The data engineer can use the Pull button in the Databricks Repos UI, or use the git pull command in a terminal session. The other options are not relevant for this task, as they either push changes to the remote repository (Push), combine

  two branches (Merge), save changes to the local repository (Commit), or create a new local repository from a remote one (Clone).

  References:

  Run Git operations on Databricks Repos

  Git pull

• Question 94:

  A company is collaborating with a partner that does not use Databricks but needs access to a large historical dataset stored in Delta format. The data engineer needs to ensure that the partner can access the data securely, without the need for them to set up an account, and with read-only access.

  How should the data be shared?

  A. Share the dataset by exporting it to a CSV file and manually transferring the file to the partner's system.
  B. Grant your partner access to your Databricks workspace and assign them full write permissions to the Delta table, enabling them to modify the dataset.
  C. Share the dataset using Unity Catalog, ensuring that both teams have full write access to the data within the same organization.
  D. Share the dataset using Delta Sharing, which allows your partner to access the data using a secure, read-only URL without requiring a Databricks account, ensuring that they cannot modify the data.
  D. Share the dataset using Delta Sharing, which allows your partner to access the data using a secure, read-only URL without requiring a Databricks account, ensuring that they cannot modify the data.

  ---

  Explanation

  Delta Sharing provides secure, read-only access to Delta tables via a sharing protocol, allowing partners without a Databricks account to query the data through a secure URL while ensuring they cannot modify it.

• Question 95:

  Which Databricks compute option automatically scales to handle query load without manual cluster management?

  A. Job Cluster
  B. All-Purpose Cluster
  C. Serverless SQL Warehouse
  D. Interactive Cluster
  E. Photon-optimized cluster
  C. Serverless SQL Warehouse

  ---

  Explanation

  Serverless SQL Warehouse provides auto-scaling compute that starts and stops automatically, optimizing cost and performance.

• Question 96:

  A data engineer is inspecting an ETL pipeline based on a Pyspark job that consistently encounters performance bottlenecks. Based on developer feedback, the data engineer assumes the job is low on compute resources. To pinpoint the issue, the data engineer observes the Spark UI and finds out the job has a high CPU time vs Task time.

  Which course of action should the data engineer take?

  A. High CPU time vs Task time means an under-utilized cluster. The data engineer may need to repartition data to spread the jobs more evenly throughout the cluster.
  B. High CPU time vs Task time means efficient use of cluster and no change needed
  C. High CPU time vs Task time means a CPU over-utilized job. The data engineer may need to consider executor and core tuning or resizing the cluster
  D. High CPU time vs Task time means over-utilized memory and the need to increase parallelism
  C. High CPU time vs Task time means a CPU over-utilized job. The data engineer may need to consider executor and core tuning or resizing the cluster

  ---

  Explanation

  A high CPU time vs Task time indicates that tasks are CPU-bound and the job is over-utilizing CPU resources. The appropriate action is to tune executors and cores or resize the cluster to ensure adequate compute resources.

• Question 97:

  A data engineer wants to create a new table containing the names of customers that live in France. They have written the following command:

  

  A senior data engineer mentions that it is organization policy to include a table property indicating that the new table includes personally identifiable information (PII). Which of the following lines of code fills in the above blank to successfully complete the task?

  A. There is no way to indicate whether a table contains PII.
  B. "COMMENT PII"
  C. TBLPROPERTIES PII
  D. COMMENT "Contains PII"
  E. PII
  D. COMMENT "Contains PII"

  ---

  Explanation

  In Databricks, when creating a table, you can add a comment to columns or the entire table to provide more information about the data it contains. In this case, since it's organization policy to indicate that the new table includes personally identifiable information (PII), option D is correct. The line of code would be added after defining the table structure and before closing with a semicolon.

  References: Data Engineer Associate Exam Guide, CREATE TABLE USING (Databricks SQL)

• Question 98:

  A data engineer wants to track the historical versions of a Delta table to analyze data as it existed one week ago. Which Delta Lake feature enables this?

  A. File history
  B. Time travel
  C. Snapshot isolation
  D. Data compaction
  E. Change data feed
  B. Time travel

  ---

  Explanation

  Delta Lake's time travel uses transaction logs to query data "as of" a timestamp or version, supporting reproducibility and debugging.

• Question 99:

  An organization has data stored across multiple external systems, including MySQL, Amazon Redshift, and Google BigQuery. The data engineer wants to perform analytics without ingesting directly into Databricks, ensuring unified governance and minimizing data duplication.

  Which feature of Databricks enables querying these external data sources while maintaining centralized governance?

  A. Delta Lake
  B. Lakehouse Federation
  C. MLflow
  D. Databricks Connect
  B. Lakehouse Federation

  ---

  Explanation

  Lakehouse Federation allows Databricks to query external data sources like MySQL, Redshift, and BigQuery without moving or duplicating data, while maintaining centralized governance through Unity Catalog.

• Question 100:

  A company sells products across multiple categories (e.g., Electronics, Clothing) and regions. The sales team has provided you with a PySpark dataframe named sales_df as below, and the team wants the data engineer to analyze the sales data to help make strategic decisions.

  

  Calculate the total sales amount for each region and store the results in a new dataframe called . region_sales

  Given the expected result:

  

  Which code will generate the expected result?

  A. region_sales = sales_df.groupBy("category").sum("sales_amount").alias ("total_sales_amount")
  B. region_sales = sales_df.groupBy("region").agg(sum("sales_amount").alias ("total_sales_amount"))
  C. region_sales = sales_df.sum("sales_amount").groupBy("region").alias ("total_sales_amount")
  D. region_sales = sales_df.agg(sum("sales_amount").groupBy("region").alias ("total_sales_amount"))
  B. region_sales = sales_df.groupBy("region").agg(sum("sales_amount").alias ("total_sales_amount"))

  ---

  Explanation

  To calculate total sales per region, you group by "region" and apply an aggregation using sum ("sales_amount"), aliasing it as total_sales_amount. This produces the expected result of 250 (North), 500 (South), 250 (East), and 400 (West).