12 real DP-900 sample questions, each with an explanation of why every option is right or wrong. No account, no card. This is the reasoning the DP-900 tests: knowing why the tempting answer is wrong, not just spotting the right one.
The real DP-900 is Typically 40 to 60 questions questions in 45 minutes, pass mark 700 / 1000. For a domain-by-domain breakdown and a study plan, read the DP-900 study guide. The full bank has 262 questions.
lock_openFree sampleDescribe Core Data Conceptsmedium
Storing customer profiles as JSON documents whose fields vary between records is an example of semi-structured data. Is this statement correct?
- AYescheck_circle Correct
- BNo
JSON documents with fields that differ between records are semi-structured, not unstructured. JSON is named in the grounding as a common semi-structured format precisely because its documents can vary in their specific fields between instances.
Why A is correct: Correct. JSON is named in the grounding as a common semi-structured format precisely because its documents can vary in their specific fields between instances.
Why B is wrong: Answering No is wrong because semi-structured data is defined as having some structure while allowing variation between instances, and JSON with varying fields is the common example the grounding gives.
lock_openFree sampleDescribe Core Data Conceptsmedium
In which type of database is structured data held in tables, with each entity instance assigned a primary key that other tables reference to eliminate duplicate values?
- AGraph database
- BRelational databasecheck_circle Correct
- CKey-value database
- DDocument database
Relational databases use tables and primary keys to reference entities and normalise data, storing each value only once. Relational databases store structured data in tables representing entities, assign each instance a primary key, and reference those keys from other tables, which enables normalisation so details such as an individual customer are stored only once.
Why A is wrong: A graph database stores entities as nodes with links defining relationships, rather than normalised tables that reference each other through primary keys.
Why B is correct: Correct. Relational databases store structured data in tables representing entities, assign each instance a primary key, and reference those keys from other tables, which enables normalisation so details such as an individual customer are stored only once.
Why C is wrong: A key-value database pairs a unique key with a value in any format, so it does not model entities as related tables joined by primary keys.
Why D is wrong: A document database stores each value as a JSON document optimised for parsing, not as normalised tables linked by primary keys to remove duplication.
lock_openFree sampleDescribe Core Data Conceptsmedium
A data engineering team needs to store large analytical datasets in a data lakehouse, where queries usually read just a few columns and benefit from strong compression and encoding. Which file format best fits this requirement?
- AComma-separated values (CSV) delimited text, kept human-readable for any application to open
- BParquet, the columnar format that is the de facto standard for modern data lakehousescheck_circle Correct
- CJSON documents, using a hierarchical schema to capture flexible attributes per record
- DXML markup, with tagged elements and attributes enclosed in angle-brackets
Parquet is a columnar, compressible format and the de facto standard for data lakehouse analytics. Parquet is a columnar format described as the de facto standard for modern data lakehouses; storing each column's data together and including chunk metadata lets an application locate and read only the needed columns while supporting efficient compression and encoding schemes.
Why A is wrong: CSV suits structured data that must stay human-readable across many tools, but it is not optimized for storage space or efficient column-based processing, so it is the wrong fit for lakehouse analytics.
Why B is correct: Correct. Parquet is a columnar format described as the de facto standard for modern data lakehouses; storing each column's data together and including chunk metadata lets an application locate and read only the needed columns while supporting efficient compression and encoding schemes.
Why C is wrong: JSON is flexible for structured and semi-structured data, but it is a human-readable text format with no columnar layout or efficient encoding, so it does not match the compressed analytical requirement.
Why D is wrong: XML is human-readable and was popular in the 1990s and 2000s, but it is verbose and not optimized for storage or column-based query, making it unsuitable here.
lock_openFree sampleDescribe an Analytics Workload on Azureeasy
In a data warehouse that uses a star schema, what is the role of the central fact table relative to the surrounding dimension tables?
- AIt holds the numeric values that can be aggregated by the related dimensionscheck_circle Correct
- BIt holds the descriptive entities used to group and slice the numeric values
- CIt stores the raw files exactly as received before any schema is applied
- DIt records the read-time formula language used to compute query measures
Fact tables hold numbers; dimension tables hold the entities to aggregate by. In a star schema the central fact table contains the numeric values, related to dimension tables that represent the entities by which the data can be aggregated.
Why A is correct: Correct. In a star schema the central fact table contains the numeric values, related to dimension tables that represent the entities by which the data can be aggregated.
Why B is wrong: Describing entities such as customer, product and store is the role of the dimension tables.
Why C is wrong: Storing raw files with no schema describes a data lake, not a fact table in a star schema.
Why D is wrong: Formula-language measures belong to a semantic model, not to a warehouse fact table.
lock_openFree sampleDescribe an Analytics Workload on Azureeasy
Which open-source distributed processing engine is Azure Databricks built on?
- AApache Spark, used for large-scale engineering, data science, and SQL analyticscheck_circle Correct
- BApache Kafka, used as a streaming event source for real-time ingestion pipelines
- CSQL Server Analysis Services, used to build pre-aggregated multidimensional cube models
- DPower Query, used to author low-code reusable data transformation logic visually
Azure Databricks is built on Apache Spark for code-first large-scale analytics. The grounding states Azure Databricks is a cloud analytics platform built on Apache Spark, optimised for large-scale data engineering, data science, and SQL analytics.
Why A is correct: Correct. The grounding states Azure Databricks is a cloud analytics platform built on Apache Spark, optimised for large-scale data engineering, data science, and SQL analytics.
Why B is wrong: Apache Kafka is named only as an event source that streaming ingestion connects to, not the engine Databricks is built on.
Why C is wrong: SQL Server Analysis Services builds multidimensional cubes for analytical models, and is unrelated to the Databricks compute engine.
Why D is wrong: Power Query underpins low-code Dataflows Gen2 transformations, and is not the processing engine behind Azure Databricks.
lock_openFree sampleDescribe an Analytics Workload on Azuremedium
Fabric Mirroring can continuously replicate an external database such as Azure SQL Database into OneLake without any pipeline authoring. Is this statement correct?
- AYescheck_circle Correct
- BNo
Fabric Mirroring auto-replicates external databases into OneLake with no pipeline authoring. Fabric Mirroring continuously replicates external databases including Azure SQL Database, Snowflake, and Azure Cosmos DB into OneLake in near-real-time, with Fabric handling change tracking and replication automatically after a one-time source connection, requiring no pipeline authoring.
Why A is correct: Correct. Fabric Mirroring continuously replicates external databases including Azure SQL Database, Snowflake, and Azure Cosmos DB into OneLake in near-real-time, with Fabric handling change tracking and replication automatically after a one-time source connection, requiring no pipeline authoring.
Why B is wrong: Answering No would wrongly imply replication needs a hand-built pipeline, but Mirroring is configured once and then managed automatically by Fabric.
lock_openFree sampleIdentify Considerations for Relational Data on Azureeasy
In a relational database, how is a single instance of a real-world entity such as one customer represented?
- AAs one row within the table that models that entitycheck_circle Correct
- BAs one column within the table that models that entity
- CAs a separate table created for that individual instance
- DAs a datatype definition applied to the whole table
A row is one instance of an entity; the table is the whole collection of those instances. A relational table contains rows where each row represents a single instance of an entity, so one customer occupies one row in the customer table.
Why A is correct: Correct. A relational table contains rows where each row represents a single instance of an entity, so one customer occupies one row in the customer table.
Why B is wrong: A column stores one attribute (such as Email) across many entity instances, not a single instance.
Why C is wrong: A table models a whole collection of entities; one instance is a row inside it, not its own table.
Why D is wrong: A datatype constrains the values a column may hold; it does not represent an entity instance.
lock_openFree sampleIdentify Considerations for Relational Data on Azureeasy
Relational tables are a format for structured data in which each row in a table has the same columns, although some columns may hold no value (NULL) for a given row. Is this statement correct?
- AYescheck_circle Correct
- BNo
Every row in a relational table shares the same columns, but a cell may be empty (NULL). Relational tables store structured data where every row shares the same set of columns, and a column such as MiddleName can be empty or NULL for rows where no value applies.
Why A is correct: Correct. Relational tables store structured data where every row shares the same set of columns, and a column such as MiddleName can be empty or NULL for rows where no value applies.
Why B is wrong: Answering No would wrongly suggest rows can have differing columns; in a relational table the columns are fixed across all rows even though individual cells may be NULL.
lock_openFree sampleIdentify Considerations for Relational Data on Azureeasy
Why does each column in a relational table store data of a specific datatype, such as decimal for a Price column or date/time for an OrderDate column?
- ASo that every column in the table is required to allow NULL values in all of its rows
- BSo that the values in the column are constrained to an appropriate datatypecheck_circle Correct
- CSo that the table can be split automatically into smaller related tables
- DSo that the same value is repeated across every row of the column
A column's datatype constrains its values to one appropriate kind of data, such as text, decimal or date/time. Each column is defined with a datatype so its values are constrained to an appropriate kind of data, such as decimal numbers for prices or date/time values for an order date.
Why A is wrong: Allowing NULL is a separate column property; it is not the purpose of assigning a datatype.
Why B is correct: Correct. Each column is defined with a datatype so its values are constrained to an appropriate kind of data, such as decimal numbers for prices or date/time values for an order date.
Why C is wrong: Splitting data into related tables is part of normalization, not a result of assigning a datatype.
Why D is wrong: A datatype governs the kind of value allowed, not whether values repeat between rows.
lock_openFree sampleDescribe Considerations for Working with Non-Relational Data on Azureeasy
A team needs to store large numbers of images, video files, and backup archives so that applications can read and write them through a cloud storage API. Which Azure storage service is designed for this kind of unstructured data?
- AAzure Blob storagecheck_circle Correct
- BAzure Files
- CAzure Table storage
- DAzure Queue storage
Blob storage is Azure's object store for unstructured data such as images, video, and backups. Azure Blob storage stores massive amounts of unstructured data as binary large objects (blobs) that applications read and write through the Blob storage API, making it the fit for images, video, and archives.
Why A is correct: Correct. Azure Blob storage stores massive amounts of unstructured data as binary large objects (blobs) that applications read and write through the Blob storage API, making it the fit for images, video, and archives.
Why B is wrong: Azure Files provides cloud network file shares mounted over SMB or NFS, not an object store accessed through a Blob API.
Why C is wrong: Table storage holds semi-structured key/value rows, not arbitrary binary objects such as images or video.
Why D is wrong: Queue storage holds messages for asynchronous processing between application components, not large binary files.
lock_openFree sampleDescribe Considerations for Working with Non-Relational Data on Azuremedium
Azure Blob storage supports several blob types. Which two statements correctly describe block blobs? (Select 2 answers)
- AA block blob is a set of blocks, and the block is the smallest unit read or written individuallycheck_circle Correct
- BBlock blobs are best suited to storing discrete, large binary objects that change infrequentlycheck_circle Correct
- CA block blob is organised as a set of fixed-size 512-byte pages optimised for random read and write
- DA block blob only permits new data to be appended to the end, and existing blocks cannot be updated
- EAzure uses block blobs to provide the virtual disk storage used by its Azure virtual machines
Block blobs are made of blocks and suit large, infrequently changed binary objects, unlike page blobs (VM disks) or append blobs (append-only). Block blobs are handled as a set of blocks where the block is the smallest individually readable or writable unit, and they are best for discrete, large binary objects that change infrequently. The other options describe page blobs or append blobs.
Why A is correct: Correct. A block blob is a set of blocks, and the block is the smallest unit read or written individually is one of the keyed answers. Block blobs are handled as a set of blocks where the block is the smallest individually readable or writable unit, and they are best for discrete, large binary objects that change infrequently.
Why B is correct: Correct. Block blobs are best suited to storing discrete, large binary objects that change infrequently is one of the keyed answers. Block blobs are handled as a set of blocks where the block is the smallest individually readable or writable unit, and they are best for discrete, large binary objects that change infrequently.
Why C is wrong: Fixed-size 512-byte pages optimised for random access describe page blobs, not block blobs.
Why D is wrong: Appending only to the end describes append blobs, where updating or deleting existing blocks is not supported.
Why E is wrong: Azure uses page blobs, not block blobs, to implement virtual disk storage for virtual machines.
lock_openFree sampleDescribe Considerations for Working with Non-Relational Data on Azuremedium
An organisation already has Azure file shares but wants to keep locally cached copies of those shared files on its on-premises servers synchronised with the data held in Azure. Which Azure Files capability provides this?
- AAzure File Synccheck_circle Correct
- BAzCopy
- CLifecycle management policy
- DGeo-redundant storage
Azure File Sync keeps locally cached copies of files in step with the share stored in Azure. Azure File Sync synchronises locally cached copies of shared files with the data in Azure File storage, giving on-premises servers an up-to-date local cache.
Why A is correct: Correct. Azure File Sync synchronises locally cached copies of shared files with the data in Azure File storage, giving on-premises servers an up-to-date local cache.
Why B is wrong: AzCopy is a utility for uploading and copying files to storage, not a service that keeps local caches continuously synchronised with the share.
Why C is wrong: Lifecycle management policies move or delete blobs across access tiers based on age; they apply to Blob storage, not file-share synchronisation.
Why D is wrong: Geo-redundant storage replicates data to a secondary region for resilience; it does not cache and sync files to on-premises servers.
Examworthy is not affiliated with or endorsed by Microsoft. All questions are original, blueprint-aligned practice material. We never reproduce live exam items. DP-900 and related marks belong to their respective owners.
