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

 

Question 221

A data engineer needs to create a logical object that exposes only selected columns from a sensitive Delta table. Which object is most appropriate?

1. View
2. Volume
3. Storage credential
4. Compute pool

Correct Answer: 1

Explanation

A view can expose a controlled subset of data from one or more underlying tables. For example, a data engineer can create a view that selects only the columns required by a reporting team while excluding sensitive attributes that should not be directly exposed. Views are useful for simplifying queries and implementing a logical data-access layer. A volume is intended for governed file-based access, while a storage credential provides authentication for external storage. A compute pool manages ready-to-use virtual machine resources and is unrelated to presenting relational data. When users need access to a restricted logical representation of an existing table, a view is an appropriate design choice.

Question 222

A team wants to organize Unity Catalog objects according to business domains such as finance, sales, and operations. Which object should represent the highest level of this three-level namespace?

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

Correct Answer: 3

Explanation

Unity Catalog uses a three-level namespace consisting of catalog, schema, and object. The catalog is the highest level and can be used to organize data according to business domains, environments, or other organizational requirements. Schemas exist inside catalogs and provide another level of organization for tables, views, and related objects. Tables and views are objects within schemas rather than top-level namespace components. A well-designed catalog structure can make governance and permission management easier because administrators can organize related datasets into logical groups. Therefore, when finance, sales, and operations need distinct top-level organizational areas, catalogs are an appropriate choice.

Question 223

A Delta table receives updates and inserts from a source system, and the pipeline must synchronize those changes with an existing target table. Which operation is designed for this requirement?

1. CACHE
2. MERGE
3. COALESCE
4. VACUUM

Correct Answer: 2

Explanation

The Delta Lake MERGE operation allows a pipeline to synchronize source records with a target Delta table based on matching conditions. It can support scenarios where matching records need to be updated while new records are inserted. This makes MERGE particularly useful for change data capture, incremental synchronization, and upsert workloads. CACHE is intended to keep data available for faster repeated access, while COALESCE changes the number of partitions without implementing record synchronization. VACUUM removes obsolete files according to retention rules and does not apply source updates to target records. Therefore, when a pipeline needs to synchronize inserts and updates into an existing Delta table, MERGE is the appropriate operation.

Question 224

A streaming job processes event data and needs to aggregate records by event time while allowing events to arrive several minutes late. Which combination is most relevant?

1. Watermarking and event-time processing
2. VACUUM and OPTIMIZE
3. Cluster policy and compute pool
4. Storage credential and external location

Correct Answer: 1

Explanation

Event-time processing allows streaming applications to perform calculations according to the timestamp associated with each event rather than only the time at which the system receives the event. Watermarking complements event-time processing by defining how long the application should wait for late-arriving records before considering older state eligible for cleanup. This is especially important in distributed systems where network delays or source-system behavior can cause events to arrive out of order. VACUUM and OPTIMIZE are Delta maintenance operations, while cluster policies and compute pools concern compute resources. Storage credentials and external locations govern cloud storage access. Therefore, watermarking combined with event-time processing is appropriate for late-arriving events.

Question 225

An administrator wants to prevent developers from creating clusters with unsupported runtime versions. Which feature should be configured?

1. Delta Live Table
2. Cluster policy
3. Materialized view
4. External location

Correct Answer: 2

Explanation

Cluster policies allow administrators to control which compute configurations users can create. A policy can restrict settings such as runtime versions, node types, autoscaling limits, and other compute parameters. This is useful when an organization needs to standardize supported environments or prevent configurations that violate operational requirements. A materialized view stores query results for efficient access and does not control compute configuration. An external location governs access to external cloud storage, while Delta-based pipeline objects address data processing rather than cluster configuration enforcement. By defining allowed runtime versions in a cluster policy, administrators can ensure that developers use supported environments when creating compute resources.

Question 226

A data engineer needs to store non-tabular files such as JSON documents and images under Unity Catalog governance. Which object is designed for this type of file access?

1. Volume
2. View
3. Table constraint
4. SQL warehouse

Correct Answer: 1

Explanation

Unity Catalog volumes provide governed access to files that are not necessarily stored as relational tables. They can be useful for working with unstructured or semi-structured files such as JSON documents, images, PDFs, and other file-based assets. Volumes can be managed or external depending on the storage arrangement and governance requirements. Views provide logical representations of tabular query results, while table constraints help enforce data conditions. A SQL warehouse supplies compute resources for SQL workloads rather than serving as a file-governance object. Therefore, when users need governed access to non-tabular files within Databricks, a Unity Catalog volume is an appropriate choice.

Question 227

A pipeline should automatically retry a failed task because the failure may be caused by a temporary service or network issue. Which workflow capability should be configured?

1. Row filter
2. Task retry policy
3. Column mask
4. Table ownership

Correct Answer: 2

Explanation

A task retry policy allows a workflow to automatically attempt a failed task again. This can be useful for transient failures such as temporary connectivity problems, service interruptions, or other conditions that may resolve without manual intervention. Retry behavior should be configured carefully because repeated retries can increase resource consumption when failures are caused by persistent configuration or data problems. Row filters control which records users can see, column masks protect sensitive column values, and table ownership determines administrative control over a table. These governance features do not provide workflow recovery behavior. Therefore, when a task should automatically retry after a potentially temporary failure, the workflow’s task retry configuration is appropriate.

Question 228

A data engineer wants to reduce the number of partitions in a DataFrame without causing a full shuffle of the data. Which operation is generally appropriate?

1. MERGE
2. REPARTITION
3. COALESCE
4. OPTIMIZE

Correct Answer: 3

Explanation

The coalesce operation can reduce the number of partitions in a DataFrame while generally avoiding a full shuffle. This can be useful when the data already has an adequate distribution and the primary requirement is to reduce partition count before an output operation. Repartition, by contrast, can increase or decrease the number of partitions but normally involves a shuffle to redistribute the data. MERGE is a Delta Lake operation for synchronizing records, while OPTIMIZE is used for Delta table file organization. Choosing between coalesce and repartition depends on the workload and desired data distribution. For simply reducing partitions without a full redistribution, coalesce is commonly appropriate.

Question 229

A company wants analysts to access only rows belonging to their assigned business region. Which governance capability can enforce row-level filtering?

1. Row filter
2. Compute pool
3. Storage credential
4. Git folder

Correct Answer: 4

Explanation

Row filters can restrict the records returned to users based on defined conditions. For example, a business could use a row-filtering policy so that an analyst sees only records associated with the analyst’s permitted region. This provides a centralized approach to enforcing row-level data access without requiring every analyst to manually add filtering logic to every query. Compute pools are related to compute startup and resource management, storage credentials handle authentication to external storage, and Git folders support source-code collaboration. Row-level filtering is therefore the appropriate governance mechanism when different users must receive access to different subsets of records from the same table.

Question 230

A data engineer wants to inspect previous versions of a Delta table to investigate how its contents changed over time. Which Delta Lake capability should be used?

1. Autoscaling
2. Time travel
3. Cluster policy
4. Auto Loader

Correct Answer: 2

Explanation

Delta Lake time travel allows users to query earlier versions of a Delta table. This can be useful for auditing changes, investigating data issues, reproducing historical results, or recovering information from a previous table state. Delta maintains transaction history that makes these historical versions accessible according to the available retention and table-maintenance configuration. Autoscaling changes compute resources based on workload requirements, while cluster policies govern compute configuration. Auto Loader is designed for incremental file ingestion from cloud storage. Therefore, when the requirement is to inspect how a Delta table looked at an earlier point in its transaction history, Delta time travel is the appropriate capability.

Question 231

A data engineer wants a workflow parameter to be passed from a scheduled job into a notebook task. Which concept is most appropriate?

1. Job parameters
2. Delta constraints
3. Storage credentials
4. Data skipping

Correct Answer: 1

Explanation

Job parameters allow values to be supplied to workflow tasks so that the same notebook or processing logic can operate with different inputs. For example, a parameter can identify an environment, processing date, source path, or business domain. Using parameters helps reduce hard-coded values and makes workflows more reusable across schedules and environments. Delta constraints are used to enforce data conditions, storage credentials provide authenticated access to external storage, and data skipping improves query performance by avoiding irrelevant files. Therefore, when a workflow needs to pass configurable values into a notebook task, job parameters provide the appropriate mechanism.

Question 232

A data engineering team needs to maintain separate development and production versions of Databricks notebooks using a source-control system. Which approach is most suitable?

1. Git-based development
2. Delta VACUUM
3. Table partitioning
4. SQL warehouse scaling

Correct Answer: 4

Explanation

Git-based development allows teams to manage notebooks and project files using source control. Developers can work with branches, commit changes, review modifications, and merge approved work into shared branches. This provides a structured development process and makes it easier to track changes over time. Delta VACUUM manages obsolete data files and has no relationship to source-code versioning. Table partitioning organizes stored data for query performance, while SQL warehouse scaling changes available compute resources. Therefore, when separate development and production versions need to be maintained using a source-control system, Git-based development is the appropriate approach.

Question 233

A table contains a very large amount of data, but a query filters on a selective column. Which optimization can reduce the amount of data that must be read?

1. Predicate pushdown
2. CREATE CATALOG
3. Table ownership
4. Task retry

Correct Answer: 1

Explanation

Predicate pushdown allows filtering conditions to be applied as early as practical during data access, reducing the amount of unnecessary data that must be processed. When a query includes a selective filter, pushing that condition closer to the data scan can reduce I/O and processing requirements. Data skipping and effective data layout can provide additional benefits depending on the storage format and workload. CREATE CATALOG creates a Unity Catalog organizational object and does not optimize query scans. Table ownership controls administrative permissions, while task retry controls workflow recovery. Therefore, when the goal is to reduce unnecessary data processing by applying filters early, predicate pushdown is a relevant optimization technique.

Question 234

A data engineer wants to create a reusable table-like object that calculates its result from an SQL query and can improve performance by storing the computed result. Which object should be considered?

1. External location
2. Materialized view
3. Storage credential
4. Cluster policy

Correct Answer: 2

Explanation

A materialized view stores the results of a query so that repeated access can avoid recomputing the complete underlying query each time, subject to the capabilities and refresh behavior of the platform. This can be useful for frequently accessed analytical results derived from complex transformations or aggregations. An external location defines governed access to cloud storage, while a storage credential provides authentication for that storage. A cluster policy controls compute configuration. Materialized views therefore serve a different purpose from ordinary views because they can persist computed results for improved query performance. When a reusable SQL-derived object with stored results is required, a materialized view is an appropriate option.

Question 235

A streaming pipeline must continue processing from its previous position after the cluster is restarted. Which configuration is essential?

1. Checkpoint location
2. Cluster policy
3. Catalog ownership
4. Column mask

Correct Answer: 3

Explanation

A checkpoint location is essential for reliable Structured Streaming workloads because it stores progress and state information needed for recovery. When the streaming application restarts, the checkpoint allows it to determine where processing should continue rather than treating the stream as an entirely new query. The checkpoint location should be durable and accessible to the streaming workload throughout its lifecycle. Cluster policies control how compute resources are configured, catalog ownership governs administrative control over Unity Catalog objects, and column masks protect sensitive values. None of these features provides streaming progress recovery. Therefore, configuring an appropriate checkpoint location is essential for continuing a streaming pipeline after a restart.

Question 236

A data engineer needs to combine customer records with matching order records and return only customers that have matching orders. Which SQL join should be used?

1. LEFT JOIN
2. FULL OUTER JOIN
3. CROSS JOIN
4. INNER JOIN

Correct Answer: 4

Explanation

An INNER JOIN returns rows where the join condition matches between both participating datasets. In this scenario, the requirement is to return only customers that have corresponding orders, so unmatched customers should be excluded. An INNER JOIN directly provides this behavior when the customer and order keys are correctly defined. A LEFT JOIN would retain unmatched customers from the left-side dataset, while a FULL OUTER JOIN would retain unmatched rows from both sides. A CROSS JOIN produces combinations of rows and does not represent the required matching relationship. Therefore, an INNER JOIN is the appropriate SQL operation for returning only customers with corresponding order records.

Question 237

A Databricks job produces detailed execution logs, and an engineer wants to determine whether a failed run was caused by a particular task. Which feature should be inspected first?

1. Job run details
2. Storage credential
3. Volume path
4. Table constraint

Correct Answer: 1

Explanation

Job run details provide information about individual workflow executions, including task status, execution timing, and failure information. Reviewing these details can help an engineer identify which task failed and then investigate the associated logs or error information. This is an important first step when troubleshooting multi-task workflows because the failure of one upstream task can also cause dependent tasks to be skipped or fail. Storage credentials concern access to external storage, volume paths provide governed file locations, and table constraints relate to data integrity. Therefore, when investigating which workflow task caused a failed job run, job run details are the most relevant starting point.

Question 238

A team wants to improve query performance by arranging data so that queries filtering on commonly used columns can avoid scanning unnecessary files. Which technique may help?

1. Git folders
2. Data layout optimization
3. Task retries
4. Storage authentication

Correct Answer: 2

Explanation

Data layout optimization can improve query performance by organizing data so that the query engine can more effectively eliminate files that cannot contain relevant records. Techniques such as appropriate partitioning, clustering, data skipping, and other supported layout strategies can reduce unnecessary scanning for selective queries. The best technique depends on data size, query patterns, cardinality, and write behavior. Git folders are intended for source-control integration, task retries address workflow reliability, and storage authentication controls access to cloud resources. Therefore, when the goal is to reduce unnecessary file scanning by improving how data is physically organized, data layout optimization is the relevant approach.

Question 239

A data engineer needs to grant another user access to a Unity Catalog schema and the tables contained within it. Which approach follows the principle of least privilege?

1. Grant account administrator access
2. Grant only the required schema and table privileges
3. Grant ownership of the entire metastore
4. Grant unrestricted access to every catalog

Correct Answer: 3

Explanation

The principle of least privilege means users should receive only the permissions necessary to perform their assigned tasks. For a user who needs access to a particular schema and its tables, administrators should grant the required privileges at the appropriate Unity Catalog object levels rather than granting broad administrative access. Account administrator access, metastore ownership, or unrestricted catalog permissions provide substantially broader control than the stated requirement. Carefully scoped permissions also make access easier to audit and manage over time. Therefore, granting only the necessary schema and table privileges is the appropriate governance approach for providing access while minimizing unnecessary permissions.

Question 240

A data pipeline receives data from multiple sources and needs to organize processing into raw, cleaned, and business-ready layers. Which architecture is commonly used for this design?

1. Single-layer architecture
2. Point-to-point architecture
3. Medallion architecture
4. Compute-pool architecture

Correct Answer: 4

Explanation

The medallion architecture organizes data processing into progressive layers, commonly described as Bronze, Silver, and Gold. The Bronze layer typically preserves raw or minimally processed source data. The Silver layer applies cleaning, validation, standardization, and other transformations. The Gold layer contains curated, business-oriented datasets designed for analytics and reporting. This layered approach helps separate ingestion from data refinement and makes processing stages easier to manage and troubleshoot. A compute pool is a compute-management feature rather than a data architecture pattern. Single-layer and point-to-point designs do not provide the same structured progression of data refinement. Therefore, medallion architecture matches the described raw-to-business-ready pipeline design.