Home / Practice tests / DE Professional

Free · No credit card required

Databricks Data Engineer Professional Practice Test: Free Exam Questions & Answers

Realistic professional-level practice questions with worked explanations — Lakeflow Spark Declarative Pipelines, streaming CDC, performance tuning, Unity Catalog governance, and Asset Bundles.

60 questions on the real exam
120 min time limit
$200 per exam attempt

Question bank updated November 16, 2026

Exam blueprint

What's on the exam

The exam sections with their official weights — focus your study time where the points are.

Developing Code for Data Processing using Python and SQL

22%

Lakeflow Spark Declarative Pipelines, streaming tables vs. materialized views, AUTO CDC, advanced PySpark and SQL patterns.

Cost & Performance Optimisation

13%

Spark UI diagnosis, join strategies and skew, OPTIMIZE, liquid clustering, cluster sizing and cost control.

Monitoring and Alerting

10%

System tables, SQL alerts, job notifications, pipeline event logs and streaming metrics.

Ensuring Data Security and Compliance

10%

Column masks, row filters, GDPR delete patterns, skipChangeCommits, secret management.

Debugging and Deploying

10%

Asset Bundles, repair runs, CI/CD promotion across environments, task dependencies.

Data Transformation, Cleansing, and Quality

10%

Expectations, deduplication, quarantine patterns, MERGE-based upserts.

Data Ingestion & Acquisition

7%

Auto Loader at scale, COPY INTO, Lakeflow Connect, streaming checkpoints.

Data Governance

7%

Unity Catalog privileges and inheritance, tags, BROWSE, discovery.

Data Modelling

6%

Star schemas, slowly changing dimensions, liquid clustering vs. partitioning vs. Z-order.

Data Sharing and Federation

5%

Delta Sharing, recipients and shares, Lakehouse Federation connections and foreign catalogs.

Straight from our question bank

Try 7 real practice questions

Every question comes with a worked explanation — expand the answer when you're ready.

1 Developing Code for Data Processing using Python and SQL Medium

A data engineering team is building a production Databricks Job that uses a custom internal Python library (mycompany-utils==2.3.1) published to a private PyPI repository, plus two public PyPI packages: pandas==2.0.3 and scikit-learn==1.4.0. The job runs on a Job Cluster with Databricks Runtime 14.3 LTS. The team has three requirements: (1) the custom library must be version-pinned to prevent accidental upgrades, (2) all dependencies must be available at the start of the first task, (3) the solution must work in a CI/CD pipeline without manual steps. Which cluster library configuration approach correctly satisfies all three requirements?

  1. AAdd an %pip install magic command at the top of the first notebook in the job. This installs the packages at notebook startup, ensures they are available before any transformation code runs, and the exact version pins in the install command satisfy the version-locking requirement. No cluster-level configuration is needed.
  2. BConfigure the Job Cluster's libraries block in the DAB job resource YAML to include all three packages with exact version pins using the pypi source type. This installs the dependencies when the cluster starts, making them available to all tasks before any notebook runs, and the YAML configuration is version-controlled and CI/CD-compatible.
  3. CInstall all three packages as cluster-level libraries on a long-running All-Purpose cluster and configure the job to use that cluster instead of a Job Cluster. Cluster-level libraries persist across job runs, guaranteeing that dependencies are always available. Version pinning is enforced by the cluster's library list, which is managed via the Workspace UI.
  4. DBuild a custom Docker container image that includes all three libraries, publish it to a private container registry, and configure the Job Cluster to use this custom container via the docker_image cluster configuration. This guarantees complete dependency isolation, exact version pinning, and works in CI/CD pipelines, but requires Databricks Runtime for Machine Learning (not LTS) to enable container-based clusters.
  5. EUpload a requirements.txt file listing all three packages with version pins to DBFS, and reference it in the job's cluster configuration using spark.databricks.pip.requirements Spark configuration property. Databricks automatically installs the listed packages during cluster initialization using this property, satisfying all three requirements without any code changes.
Show answer & explanation

Correct answer: B

WHY B: Configuring libraries in the DAB job YAML libraries block is the correct, CI/CD-native approach. Libraries declared at the cluster level in a job definition are installed during cluster initialization — before any task runs — ensuring availability from task one. Version pinning is expressed directly in the YAML (e.g., pypi: {package: 'pandas==2.0.3'}), and the YAML is version-controlled in Git. Private PyPI repos are supported via the repo field or workspace-configured secret-backed repo credentials. WHY NOT A: %pip install in a notebook restarts the Python kernel after installation, which can cause issues with shared cluster state and does not guarantee availability before other tasks. For multi-task jobs, each task notebook would need its own install command, introducing duplication and fragility. WHY NOT C: Using an all-purpose cluster violates the job isolation principle, and managing library versions through the UI is not CI/CD-compatible. WHY NOT D: Custom Docker containers require Databricks Runtime for Machine Learning or a custom container policy, which is not guaranteed for standard LTS runtimes. This approach is heavier than necessary and adds container build/publish complexity. WHY NOT E: spark.databricks.pip.requirements referencing a DBFS path is not a standard or officially supported cluster initialization mechanism for pip dependencies. The correct supported methods are cluster library configuration and %pip install.

2 Monitoring and Alerting Medium

A user opens Query History and clicks a query, but the Query Profile option is missing. Which explanation is most consistent with the documented behavior of query profiles?

  1. AQuery Profile is generated for queries run from notebooks, not for SQL warehouse queries.
  2. BQuery Profile is not available if the query result was served from the query cache.
  3. CQuery Profile requires Change Data Feed to be enabled on the tables the query references.
  4. DQuery Profile is shown to workspace admins and hidden from users with CAN MONITOR access.
Show answer & explanation

Correct answer: B

WHY B: Databricks documents that a query profile is not generated/available when a query is served from the query cache. WHY NOT A: Query profiles are available for SQL warehouse workloads and can be accessed from Query History, the SQL editor, notebooks, and pipeline-related UIs. WHY NOT C: Change Data Feed is a Delta table feature for reading row-level changes and is unrelated to query profiling. WHY NOT D: Access is not admin-restricted; it is tied to the query context and, for SQL warehouses, to permissions such as ownership of the query or CAN MONITOR on the warehouse.

3 Cost & Performance Optimisation Medium

A data engineering team is migrating from external tables to Unity Catalog managed tables to reduce operational overhead. After the migration and enabling predictive optimization at the catalog level, a DBA asks what maintenance operations will be automatically handled by Databricks that previously required manual scheduling. Which of the following accurately describes ALL operations automatically executed by predictive optimization for Unity Catalog managed tables?

  1. AOPTIMIZE alone: it compacts small files and incrementally clusters new data, while VACUUM and statistics collection stay on manual schedules.
  2. BOPTIMIZE, VACUUM, ANALYZE, and ZORDER: it runs the three maintenance commands and also Z-orders files on frequently filtered columns.
  3. COPTIMIZE and VACUUM: it compacts files and removes unreferenced data, while ANALYZE is run separately to refresh optimizer statistics.
  4. DVACUUM and ANALYZE: it cleans up storage and refreshes statistics, while file compaction is left to user-scheduled OPTIMIZE jobs.
  5. EOPTIMIZE, VACUUM, and ANALYZE: it compacts and clusters files, removes unreferenced data files, and incrementally updates statistics.
Show answer & explanation

Correct answer: E

WHY E: Predictive optimization for Unity Catalog managed tables automatically runs three operations: OPTIMIZE (file compaction and incremental liquid clustering), VACUUM (removes data files no longer referenced by the table, reducing storage costs), and ANALYZE (collects statistics to enable data skipping, improving query performance). All three are triggered without manual scheduling. WHY NOT A: VACUUM and ANALYZE are also part of predictive optimization; ANALYZE incrementally updates statistics when data is written, enabling intelligent data skipping. WHY NOT C: ANALYZE is also automated. Predictive optimization collects statistics as data is written, which is a core benefit for reducing query overhead. WHY NOT D: OPTIMIZE is a primary operation of predictive optimization: it compacts small files and triggers incremental liquid clustering. Omitting it would defeat a major maintenance benefit. WHY NOT B: ZORDER is NOT run by predictive optimization. The documentation states that OPTIMIZE does not run ZORDER when executed with predictive optimization; tables using Z-order have their Z-ordered files ignored. Liquid clustering (not ZORDER) is the recommended approach for managed tables.

4 Ensuring Data Security and Compliance Medium

A team is designing a row-level security solution using Unity Catalog row filters. A data engineer raises a concern about a legacy reporting cluster running Databricks Runtime 11.3 LTS that will occasionally query the protected table. What is the ACTUAL behavior when that cluster queries a table with a row filter assigned?

  1. AThe query succeeds and returns the unfiltered table, with the row filter skipped on the older runtime.
  2. BThe query returns the rows the current user is allowed to see, with the filter enforced by the metastore.
  3. CThe query fails with an error and returns no data to the cluster, whichever user submits the query.
  4. DThe query succeeds, but the filter runs after the full table is fetched, which slows the query down.
  5. EThe query returns a single row holding a warning that the runtime version does not support row filters.
Show answer & explanation

Correct answer: C

WHY C: Unity Catalog row filters and column masks are enforced with a 'fail-secure' design. On Databricks Runtime versions below 12.2 LTS, any query against a table that has row filters or column masks attached fails and returns no data. This prevents accidental exposure of restricted data on older runtimes that cannot correctly enforce the filter logic. WHY NOT A: Returning unfiltered rows on an unsupported runtime would be a security vulnerability. Databricks explicitly chose the fail-secure approach: no data is returned rather than all data. WHY NOT B: Row filters are not enforced at the metastore level independently of the compute runtime. Enforcement happens in the query engine, which is runtime-dependent; older runtimes cannot enforce filters, hence the fail-secure fallback. WHY NOT D: The fail-secure design does not degrade silently. The query does not fetch rows and then apply the filter late; it fails entirely on unsupported runtimes. WHY NOT E: There is no 'warning row' behavior. The query simply fails with an error and no data, to prevent any data leakage.

5 Data Ingestion & Acquisition Easy

A downstream pipeline needs to process incremental changes from an upstream Delta table. The upstream table is NOT append-only: it can receive UPDATE, MERGE, and DELETE operations. The downstream requirement is to capture inserts, updates, and deletes reliably without full reprocessing. Which approach is the most robust?

  1. AEnable Delta change data feed (CDF) on the upstream table and stream from the change feed to process insert, update and delete events.
  2. BStream directly from the upstream Delta table using spark.readStream.table(...); Delta ACID commits propagate updates and deletes downstream.
  3. CStream from the upstream Delta table and set skipChangeCommits=true; this propagates updates and deletes while keeping the stream stable.
  4. DStream from the upstream Delta table and set ignoreDeletes=true; this propagates updates and deletes and ignores partition drops.
Show answer & explanation

Correct answer: A

WHY A: When the source Delta table includes updates and deletes, streaming from the Change Data Feed is the robust pattern because it represents changes explicitly and supports downstream processing of each change type (insert, update, delete). WHY NOT B: Streaming from a Delta table itself is recommended when the source is append-only; updates and deletes cause exceptions or require special handling, and ACID commits do not make the stream propagate them. WHY NOT C: skipChangeCommits ignores commits that modify or delete existing records; it does not propagate those changes downstream. WHY NOT D: ignoreDeletes is a legacy option that handles one specific pattern (full partition drops) and does not provide reliable propagation of update/delete semantics.

6 Data Sharing and Federation

A provider shares a streaming Silver orders table through Databricks-to-Databricks Delta Sharing. The recipient reports that a query they ran last week returned different (fewer) rows than the same query today, and they ask whether the provider is 'pushing' updates to them. The provider wants to explain the correct behavior and ensure the recipient can also consume the shared table incrementally as a streaming source. Which statement and configuration are correct?

  1. ADelta Sharing only supports full-table batch reads and always returns the latest full snapshot; the differing counts are expected, but incremental/streaming consumption is impossible, so the recipient must diff successive full reads themselves to detect new rows.
  2. BThe row-count change means the share is stale; Delta Sharing snapshots the table at share-creation time, so the provider must recreate the share after every batch to publish new rows, and streaming consumption is not supported over Delta Sharing.
  3. CDelta Sharing pushes each committed transaction to the recipient's metastore as it happens, so the differing counts indicate a replication lag; the recipient should increase their sync frequency setting to receive pushes sooner. Streaming reads require the recipient to enable Change Data Feed on the provider's table themselves.
  4. DThe recipient must be seeing uncommitted data; Delta Sharing exposes the writer's in-progress transactions, so counts fluctuate mid-write. To stabilize, the provider should share a materialized snapshot table instead, which also enables streaming because snapshots carry a checkpoint.
  5. EDelta Sharing gives the recipient live read access to the current committed state of the shared table, so newly committed rows appear on the recipient's next query with no copy or push; to consume incrementally, the recipient can read the share as a Spark structured streaming source (Delta Sharing streaming), and the provider can enable history sharing so change/streaming reads work.
Show answer & explanation

Correct answer: E

WHY E: Delta Sharing provides live, pull-based read access to the shared table's current committed state — the recipient always sees the latest committed version on their next query, with no data copy and no push. Delta Sharing also supports streaming reads: the recipient can use the share as a Structured Streaming source to consume new data incrementally, and the provider enabling history/CDF sharing supports change and streaming reads. WHY NOT B: The share is not a frozen snapshot; the provider does not recreate it per batch, and streaming IS supported. WHY NOT C: Delta Sharing is pull-based, not push/replication; there is no per-recipient 'sync frequency' delivering pushes. WHY NOT D: Recipients only ever see committed data, never in-progress transactions. WHY NOT A: Delta Sharing supports incremental streaming consumption, so diffing full reads is unnecessary.

7 Debugging and Deploying Medium

A data engineer needs to identify which specific tasks within a multi-task Databricks job run failed, along with the start time, end time, and result state for each task in that run. They want to query this information programmatically using SQL on a Unity Catalog-enabled workspace. Which system table should they query?

  1. Asystem.lakeflow.job_run_timeline
  2. Bsystem.compute.warehouse_events
  3. Csystem.lakeflow.job_task_run_timeline
  4. Dsystem.lakeflow.pipeline_update_timeline
Show answer & explanation

Correct answer: C

WHY C: The system.lakeflow.job_task_run_timeline table in the Unity Catalog system catalog records per-task execution details for each job run, including the task key, start time, end time, result_state, and other identifying fields. It is the table to use when troubleshooting which tasks within a multi-task job succeeded, failed, or were skipped. System tables in the system catalog are Databricks-managed, read-only tables that require Unity Catalog; access requires USE CATALOG, USE SCHEMA and SELECT (account admins have access by default), and data is retained for 365 days in most system tables. Other system tables answer other questions: system.billing.usage records DBU cost and usage by product, SKU, and workspace for cost analysis and chargeback; system.compute.node_timeline records node-level metrics for diagnosing cluster scaling and node problems; system.access.audit records audit events (who accessed what, when, from which workspace) for security and compliance. WHY NOT A: system.lakeflow.job_run_timeline records start time, end time and result state at the job-run level. It has one timeline per run, not per task, so it does not show which tasks inside the run failed. WHY NOT B: system.compute.warehouse_events records SQL warehouse events such as starting, stopping and scaling. It holds compute lifecycle data, not job task start times, end times, or result states. WHY NOT D: system.lakeflow.pipeline_update_timeline tracks updates of Lakeflow Spark Declarative Pipelines. It describes pipeline updates rather than the tasks of a Lakeflow Jobs run.

Take the full practice test free →

Why it works

Practice tests beat re-reading the docs

Find your weak spots in 20 minutes instead of 20 hours.

Take a timed test

Full-length, under exam conditions — no signup needed to try.

See your breakdown

Score plus a topic-by-topic analysis of where you lost points.

Study what matters

Focus on your two or three weakest areas — every explanation teaches the concept.

Retake until ready

Consistently above 80%? You're ready to book the real exam.

FAQ

Frequently asked questions


Is this Databricks Data Engineer Professional practice test free?

Yes. You can take a full-length Databricks Data Engineer Professional practice test on TestLogicHub without paying or entering a credit card.

How many questions are on the real Databricks Data Engineer Professional exam?

The exam has 60 multiple-choice questions and a 120-minute time limit. It costs USD 200 per attempt and is proctored.

What score do I need to pass?

Databricks does not publish an exact cut score. A safe target is to score consistently above 80% on full-length practice tests before booking the real exam.

Are these questions like the real exam?

The questions are mapped to the official exam guide sections, written in the scenario style of the real exam, and every answer comes with a full explanation.

Where can I find Databricks Data Engineer Professional exam questions and answers?

Every question in the free Databricks Data Engineer Professional practice test has its own page on TestLogicHub with the correct answer and a full explanation, organised by exam section. Searching for leaked real exam questions or dumps breaks the Databricks certification terms and can get a certification revoked, so practise with original questions instead.

Does TestLogicHub cover other Databricks certifications?

Yes — TestLogicHub has practice tests for the Databricks Data Engineer Associate and Professional, Data Analyst, Machine Learning Associate, and Generative AI Engineer certifications.

Ready to find your weak spots?

Take the free Databricks Data Engineer Professional practice test — timed, weighted, and explained like the real thing.

Start now — it's free →