Free DP-750 Practice Test – 20 Questions with Explanations

A claims analytics team already has SELECT on the insurance.prod.claims table. Analysts must see only rows for their assigned region, and the member_ssn column must be obfuscated in query results, while non-sensitive columns remain readable.
Which two actions should you implement? (Select TWO.)

A CI/CD runner uses a Microsoft Entra ID service principal to call the Databricks CLI against a workspace and validate Unity Catalog grants after deployment. The authentication attempt fails during workspace-level operations even though the tenant ID, client ID, and client secret are correct.

Which change should be made?

A Lakeflow Spark Declarative Pipelines bronze-to-silver flow ingests JSON with Auto Loader. A new nullable source column appears unexpectedly, and some records now contain negative amount values that must not reach silver, but the team still wants metrics on rejected rows and wants schema changes handled with minimal manual tracking.
What is the best remediation?

A silver-to-gold pipeline has become much slower after recent transformation changes. The engineer opens the Spark UI and query profile to determine whether the main bottleneck is shuffle pressure combined with spill rather than poor caching reuse.
Which two findings most strongly support that conclusion? (Select TWO.)

An operations team needs centralized Azure Databricks workspace logs in Azure Monitor, then wants an alert when a log query identifies a recurring failure pattern. They want the cleanest end-to-end order from data collection through alerting.
Steps
1. Create a log query and alert rule in Azure Monitor
2. Configure diagnostic settings on the Databricks workspace
3. Validate that the logs arrive in a Log Analytics workspace
4. Create or identify the target Log Analytics workspace
What is the correct order?

A team wants a read-only mirror of an external PostgreSQL operational database inside Unity Catalog. A platform engineer creates the connection successfully, but another engineer who can inspect the connection still cannot create the foreign catalog.

Which privilege is required next to implement the foreign catalog?

An interactive analytics team keeps a shared all-purpose compute resource available during business hours because analysts reconnect frequently. Demand ranges from a few users to dozens of users across the day, and the team wants better right-sizing while preserving fast starts.

Which change should be made first?

A regulated analytics team keeps a Unity Catalog managed Delta table that supports rollback investigations for 30 days. The table is updated by long-running MERGE jobs, and the team also wants storage cleaned up after the rollback window without risking active writes.
Which action is the best fit?

A data engineering team is standardizing table creation in Unity Catalog. The curated sales table must be fully managed by Unity Catalog, while the landing orders table must keep using an existing cloud storage path because another platform also writes files there.

Which change is required?

A nightly pipeline rebuilds one transient table from scratch and the engineering lead wants the reset step to remove both metadata and stored data. At the same time, shared datasets and retained curated tables must not lose underlying files if a table object is dropped.

Which table is the best candidate for a DROP TABLE step that should also clean up table data?

A bronze feed contains inserts, updates, and deletes for customer records, and late-arriving changes are possible. The silver target must preserve full attribute history for analytical replay rather than only the latest state.

What is the main benefit of this approach for the requirement?

A manufacturing company must ingest files from an on-premises network and then apply custom PySpark parsing before loading curated Delta tables in Unity Catalog. The solution must support hybrid connectivity and central orchestration across non-Databricks steps.
Which TWO tool choices are the most appropriate? (Select TWO.)

A business analytics team created a Genie space for finance users. Users keep asking questions such as “gross margin by channel excluding returns,” but Genie answers are inconsistent because the business uses several internal terms that do not appear clearly in table names.

What is the best improvement to the Genie configuration?

A workspace admin created a strict production compute policy for batch jobs. A data engineer opens a compute resource that uses that policy and expects to add another library manually from the Libraries tab.

Based on the exhibit, what should the engineer expect?

A platform team must provide an external data product to partner organizations while keeping a queryable governance trail inside Databricks. One partner consumes from a non-Databricks platform, and security requires a controlled recipient model instead of handing out broad workspace credentials.
Which two actions best satisfy the requirements? (Select TWO.)

A data platform team already has several Azure Monitor alert rules covering an Azure Databricks production resource group. They want to stop paging the on-call team during a recurring Sunday maintenance window, but they do not want to disable or rewrite each alert rule individually.

Which change best meets the requirement?

An engineer is standardizing the layout of a large Delta table in Unity Catalog. The team wants a modern clustering design, but the current DDL mixes several layout mechanisms together.

Which change best corrects the design?

A data engineer plans to remove the promo_code column from a silver table in production. Before approving the change, the team wants to identify the accountable owner and check whether downstream assets depend on that column.

Which action is the best next step?

A team deploys a Lakeflow Jobs workflow through a bundle. The gold publish notebook must run only after the validation notebook has completed successfully, but the current configuration does not enforce that ordering.

What is the best correction?

A daily transformation job has one slow stage that is delaying the SLA even though most tasks finish quickly. The engineer wants to identify the stage that most strongly suggests skew rather than simply general spill or balanced heavy work.

Which stage should be investigated first for data skew?