View Full Microsoft DP-750 Exam Dumps and Practice Test Dumps.

 

Question 161

Which Azure Databricks capability provides a centralized governance model for catalogs, schemas, tables, views, and volumes?

1. Unity Catalog
2. Spark UI
3. Git folders
4. Cluster logs

Correct Answer: 1

Explanation

Unity Catalog provides centralized governance and access control for data assets across Azure Databricks. It supports a consistent permission model for catalogs, schemas, tables, views, volumes, external locations, and other supported securable objects. Administrators can assign privileges to users, groups, and service principals according to organizational requirements. Unity Catalog also supports data discovery and auditing capabilities that help organizations manage governed data environments. Spark UI is primarily used to inspect Spark execution, Git folders support source-control workflows, and cluster logs provide operational information. Therefore, Unity Catalog is the appropriate capability when centralized governance and access management are required across multiple data assets.

Question 162

Which Delta Lake operation can conditionally update, insert, or delete records based on whether source and target rows match?

1. VACUUM
2. OPTIMIZE
3. MERGE
4. DESCRIBE

Correct Answer: 3

Explanation

MERGE provides conditional synchronization between a source dataset and a target Delta table. It evaluates a specified matching condition and can execute different actions depending on the result. Matching records can be updated, unmatched records can be inserted, and appropriate conditions can also support deletes. This makes MERGE useful for change data capture, incremental processing, and maintaining current-state datasets. VACUUM removes obsolete files, OPTIMIZE reorganizes physical data files, and DESCRIBE provides metadata information. Therefore, MERGE is the appropriate Delta Lake operation when a pipeline needs to apply multiple conditional changes to a target table based on source data.

Question 163

Which Unity Catalog object is used to provide governed access to files stored in cloud storage?

1. Table
2. Volume
3. View
4. Materialized view

Correct Answer: 2

Explanation

A volume provides governed access to files through Unity Catalog. It is useful for non-tabular data such as CSV files, JSON documents, images, machine-learning artifacts, and other file-based content. Volumes allow organizations to apply centralized permissions and governance to files while keeping those assets within the Databricks data-management framework. A table represents structured tabular data, a view stores reusable query logic, and a materialized view maintains query-derived results. Therefore, a volume is the appropriate Unity Catalog object when users need controlled access to files rather than only relational table data.

Question 164

Which Structured Streaming feature helps recover a query’s progress after a failure?

1. Watermark
2. Schema evolution
3. Checkpoint
4. Partition pruning

Correct Answer: 3

Explanation

A checkpoint stores progress and state information associated with a Structured Streaming query. If the streaming workload stops because of a failure or is restarted, the checkpoint allows processing to resume using the previously recorded state rather than starting from the beginning. This is especially important for reliable streaming pipelines that use stateful transformations or require consistent incremental processing. Watermarks manage late-arriving events and state retention, schema evolution handles changes in data structure, and partition pruning is a query-performance optimization. Therefore, checkpointing is the feature specifically intended to preserve streaming progress and support recovery after interruptions.

Question 165

Which architecture pattern uses Bronze, Silver, and Gold layers for progressive data refinement?

1. Medallion architecture
2. Star schema
3. Client-server architecture
4. Hub-and-spoke architecture

Correct Answer: 1

Explanation

The medallion architecture organizes data into progressively refined layers, typically called Bronze, Silver, and Gold. Bronze commonly contains raw or minimally processed source data. Silver applies cleaning, validation, standardization, and deduplication to create reliable datasets. Gold contains curated and business-ready data intended for analytics, reporting, and downstream consumption. This layered approach helps separate ingestion from transformation and business use while improving maintainability and traceability. A star schema is a dimensional modeling pattern, while client-server and hub-and-spoke describe different architectural concepts. Therefore, the medallion architecture is the appropriate pattern for progressively refining raw data into trusted analytical datasets.

Question 166

Which Unity Catalog privilege is used to access a schema namespace?

1. SELECT
2. MODIFY
3. CREATE TABLE
4. USE SCHEMA

Correct Answer: 4

Explanation

USE SCHEMA provides access to a schema namespace in Unity Catalog. A user generally needs this privilege in addition to the appropriate catalog-level access and privileges on the specific object being accessed. SELECT provides permission to read data, MODIFY allows supported data changes, and CREATE TABLE allows table creation when the necessary permissions are available. Unity Catalog uses hierarchical access controls, meaning permissions at different levels work together rather than independently. Therefore, when the requirement is specifically to access a schema namespace so that its contained objects can be used, USE SCHEMA is the relevant privilege.

Question 167

Which feature allows sensitive values in a column to be transformed or hidden for users without sufficient access?

1. Partition pruning
2. Column masking
3. Auto Loader
4. File compaction

Correct Answer: 2

Explanation

Column masking provides fine-grained protection for sensitive values stored in columns. A masking policy can return the original value to authorized users while returning a transformed or obscured value to users who should not have direct access to the sensitive information. This approach allows organizations to maintain a shared dataset while applying different visibility rules based on authorization. Partition pruning and file compaction are primarily performance-related techniques, while Auto Loader handles incremental file ingestion. Therefore, column masking is the appropriate feature when the requirement is to protect sensitive column values without necessarily removing the underlying records from the table.

Question 168

Which operation can reduce the number of small files in a Delta table by reorganizing its physical data layout?

1. MERGE
2. VACUUM
3. OPTIMIZE
4. GRANT

Correct Answer: 3

Explanation

OPTIMIZE improves the physical organization of Delta table data and can compact many small files into fewer, larger files. Small-file accumulation is common in workloads that perform frequent incremental writes or many small streaming batches. Excessive numbers of small files can increase file-management overhead and negatively affect query performance. OPTIMIZE addresses this physical layout problem while preserving the logical contents of the table. MERGE changes records, VACUUM removes obsolete files, and GRANT manages permissions. Therefore, OPTIMIZE is the appropriate operation when the primary goal is to compact small files and improve the physical organization of a Delta table.

Question 169

Which feature is designed to detect and incrementally process new files arriving in cloud storage?

1. Auto Loader
2. Time travel
3. Cluster policy
4. Materialized view

Correct Answer: 1

Explanation

Auto Loader provides scalable incremental file ingestion from cloud storage. It tracks discovered files and processes newly arriving files without requiring the pipeline to repeatedly process the entire directory. This makes Auto Loader particularly useful for continuously growing file-based sources and large-scale ingestion workloads. It can also integrate with Structured Streaming and provide supported schema inference and schema evolution capabilities. Time travel provides access to historical Delta table versions, cluster policies govern compute configuration, and materialized views maintain query-derived results. Therefore, Auto Loader is the appropriate feature when a data engineering pipeline needs to discover and process new files as they arrive.

Question 170

Which Delta Lake feature allows users to query historical versions of a table?

1. Schema evolution
2. Time travel
3. Data skipping
4. Auto Loader

Correct Answer: 2

Explanation

Delta Lake time travel allows users to access earlier versions of a table by specifying a historical version or timestamp, subject to the availability of transaction history and required data files. It is useful for auditing, troubleshooting, reproducing historical analyses, and investigating how data changed over time. The Delta transaction log records changes to the table and supports this historical access. Schema evolution handles structural changes, data skipping helps improve query performance, and Auto Loader provides incremental file ingestion. Therefore, time travel is the appropriate Delta Lake capability when a data engineer needs to inspect or query a previous state of a table.

Question 171

Which feature helps a streaming workload manage records that arrive after their expected event time?

1. Watermark
2. Catalog
3. Volume
4. External location

Correct Answer: 1

Explanation

A watermark helps Structured Streaming manage late-arriving records by establishing an event-time threshold. It allows the streaming engine to determine how long state should be retained while waiting for delayed events. This is especially useful for stateful operations such as aggregations, joins, and deduplication. Once the watermark progresses sufficiently, older state can eventually be removed according to the semantics of the operation, helping control memory and processing requirements. Catalogs organize data namespaces, volumes provide governed file access, and external locations represent governed cloud storage paths. Therefore, watermarking is the appropriate feature for managing late-arriving events in streaming workloads.

Question 172

Which Databricks capability allows multiple tasks to be executed according to defined dependencies and schedules?

1. Unity Catalog
2. Delta Sharing
3. Lakeflow Jobs
4. External location

Correct Answer: 3

Explanation

Lakeflow Jobs provides workflow orchestration capabilities for Databricks workloads. Data engineers can define multiple tasks, establish dependencies, configure schedules or triggers, and specify retry behavior. This allows complex data pipelines to execute automatically in a controlled sequence. For example, an ingestion task can complete before a transformation task begins, followed by validation or reporting tasks. Unity Catalog provides data governance, Delta Sharing enables secure data sharing, and external locations govern cloud storage paths. Therefore, Lakeflow Jobs is the appropriate Databricks capability for orchestrating multiple data engineering tasks and defining how those tasks should execute.

Question 173

Which approach is most appropriate for preventing duplicate target records when a pipeline is rerun?

1. Increase cluster memory
2. Use idempotent processing
3. Disable Delta transactions
4. Remove data-quality checks

Correct Answer: 2

Explanation

Idempotent processing allows a pipeline to be executed repeatedly without creating unintended duplicate effects. A common design uses stable business keys, deduplication logic, MERGE operations, and appropriate state or checkpoint management. This becomes especially important when a job retries after failure or when source systems may deliver the same records more than once. Increasing cluster memory addresses resource capacity but does not prevent duplicates. Disabling Delta transactions reduces reliability, while removing data-quality checks can allow invalid or duplicated data into downstream datasets. Therefore, idempotent processing is the appropriate design principle for ensuring that rerunning a workload does not unintentionally create duplicate business records.

Question 174

Which operation removes obsolete files from a Delta table according to its retention configuration?

1. VACUUM
2. MERGE
3. OPTIMIZE
4. SELECT

Correct Answer: 1

Explanation

VACUUM removes obsolete data files that are no longer required by a Delta table according to its configured retention rules. It is a storage-maintenance operation that can help reduce unnecessary cloud-storage usage. VACUUM should be used carefully because older files may be required for historical table versions and time-travel queries. MERGE modifies records based on matching conditions, OPTIMIZE reorganizes and compacts files, and SELECT reads data. Therefore, VACUUM is the appropriate operation when a data engineer needs to clean up obsolete Delta files while following the table’s retention configuration.

Question 175

Which Unity Catalog object provides the highest level in the standard catalog-schema-object namespace?

1. Schema
2. Table
3. Catalog
4. Volume

Correct Answer: 3

Explanation

A catalog is the highest level in the standard Unity Catalog namespace hierarchy. The hierarchy generally consists of catalog, schema, and object. A catalog can contain multiple schemas, while schemas can contain tables, views, volumes, and other supported objects. This hierarchy provides organizations with a structured way to organize and govern data assets according to business domains, environments, or organizational boundaries. A schema is a level below the catalog, while tables and volumes are objects contained within schemas. Therefore, the catalog is the appropriate object when referring to the highest level of the Unity Catalog namespace.

Question 176

Which Spark operation can explicitly change the number of partitions while redistributing data across the cluster?

1. CACHE
2. VACUUM
3. REPARTITION
4. GRANT

Correct Answer: 3

Explanation

REPARTITION redistributes records across Spark partitions and allows the number of partitions to be explicitly specified. Because repartitioning generally involves a shuffle, records may move between executors, which can introduce additional processing and network overhead. It can nevertheless be useful when a workload needs greater parallelism or a different partition distribution for subsequent transformations. CACHE stores computed results for reuse, VACUUM removes obsolete Delta files, and GRANT manages access permissions. Therefore, REPARTITION is the appropriate Spark operation when a data engineer needs to change partition count while redistributing the dataset across the cluster.

Question 177

Which capability can apply custom processing logic to each micro-batch produced by a Structured Streaming query?

1. foreachBatch
2. OPTIMIZE
3. SHOW GRANTS
4. USE CATALOG

Correct Answer: 4

Explanation

foreachBatch allows custom processing logic to run for each micro-batch generated by a Structured Streaming query. It is useful when a streaming workload needs to perform operations that are easier to express using batch-oriented APIs. For example, a pipeline can use foreachBatch to execute custom transformations or apply a Delta MERGE for each micro-batch. OPTIMIZE manages Delta table file organization, SHOW GRANTS displays Unity Catalog permissions, and USE CATALOG changes the active catalog namespace. Therefore, foreachBatch is the appropriate feature when custom processing needs to be applied independently to every micro-batch of a streaming workload.

Question 178

Which feature helps administrators control allowed compute configurations for users in Azure Databricks?

1. Auto Loader
2. Cluster policy
3. Time travel
4. Materialized view

Correct Answer: 2

Explanation

Cluster policies allow administrators to define and restrict configuration settings for Databricks compute resources. Policies can control supported settings such as runtime versions, node types, autoscaling limits, and other compute parameters. This helps organizations standardize resource configurations and enforce operational or governance requirements. Auto Loader is used for incremental file ingestion, time travel provides historical Delta table access, and materialized views maintain query-derived results. Cluster policies therefore address compute governance rather than data ingestion or query storage. They are useful when administrators need to prevent unsuitable configurations and provide users with controlled options for creating Databricks compute resources.

Question 179

Which Unity Catalog capability can be used to inspect privileges assigned to a securable object?

1. MERGE
2. SHOW GRANTS
3. OPTIMIZE
4. VACUUM

Correct Answer: 2

Explanation

SHOW GRANTS provides information about privileges assigned to supported Unity Catalog securable objects. Data engineers and administrators can use it to inspect which users, groups, or service principals have been granted particular permissions. This is useful for troubleshooting authorization issues, verifying governance configurations, and reviewing access assignments. MERGE modifies Delta table records, OPTIMIZE reorganizes Delta data files, and VACUUM removes obsolete files. Therefore, SHOW GRANTS is the appropriate capability when the requirement is to inspect existing permissions on a Unity Catalog object and understand which principals have access.

Question 180

Which technique allows a query engine to skip unnecessary partitions when filtering on a partitioning column?

1. Column masking
2. Schema evolution
3. Partition pruning
4. Time travel

Correct Answer: 3

Explanation

Partition pruning allows the query engine to avoid scanning partitions that cannot contain records matching the query filter. For example, if a Delta table is partitioned by date and a query filters for a specific date, unrelated date partitions can potentially be skipped. This reduces the amount of data read and processed, which can improve query performance and reduce unnecessary I/O. Column masking controls sensitive values, schema evolution handles changes to data structures, and time travel provides access to historical table versions. Therefore, partition pruning is the appropriate optimization technique when query filters align with the table’s partitioning strategy.