DP-900: Azure Data Fundamentals Practice Exam

What the DP-900 exam covers

• Describe Core Data Concepts199 questions
• Identify Considerations for Relational Data177 questions
• Describe Considerations for Non-Relational Data137 questions
• Describe an Analytics Workload204 questions

Free DP-900 practice test questions

A sample of 10 questions with answers and explanations. Sign up free to practice all 717.

1. Question 1Describe Core Data Concepts

   A logistics company tracks millions of package deliveries per day. Each delivery record includes a tracking number, sender address, recipient address, weight, and delivery status. The operations team needs to update delivery statuses in real time as packages move through sorting facilities. Which type of data processing workload best describes this system?

   • AOnline Transaction Processing (OLTP)Correct
   • BBatch processing
   • COnline Analytical Processing (OLAP)
   • DData warehousing
   ✓ Correct answer: A

   OLTP systems are specifically designed to handle frequent, short-duration transactions where individual records are inserted, updated, or retrieved in real time. The logistics company needs to update package statuses as they change at sorting facilities, which is a classic example of real-time transactional workload. OLTP systems prioritize write and update performance, maintain referential integrity, and ensure data consistency for concurrent users, making them ideal for this delivery tracking scenario.

   Why the other options are wrong

   • BBatch processing is incorrect because it handles large volumes of data in scheduled groups, not in real-time updates as packages move through facilities.
   • COnline Analytical Processing (OLAP) is incorrect because it is optimized for complex aggregate queries and historical analysis, not for frequent updates to individual delivery records.
   • DData warehousing is incorrect because it consolidates data for analytical reporting, not for real-time operational updates.
2. Question 2Describe Core Data Concepts

   A data engineering team is designing a data pipeline for a bank. They need to extract transaction records from the core banking system, perform currency conversions, anonymize customer names, and validate data quality before loading it into a data warehouse. In which pipeline architecture are these transformation steps performed before loading data into the target?

   • AETL (Extract, Transform, Load)Correct
   • BCDC (Change Data Capture)
   • CELT (Extract, Load, Transform)
   • DMPP (Massively Parallel Processing)
   ✓ Correct answer: A

   ETL architecture extracts data from source systems, applies transformations (currency conversion, anonymization, validation) in a separate processing stage, and then loads the clean, transformed data into the target data warehouse. The key characteristic is that transformations occur before loading, ensuring data quality and consistency rules are enforced before data reaches the warehouse. For banking scenarios requiring data security and quality assurance, ETL provides the control and validation necessary.

   Why the other options are wrong

   • BCDC (Change Data Capture) is incorrect because it is a technique for identifying and replicating only changed data from source systems, not a complete architecture pattern encompassing extraction, transformation, and loading.
   • CELT (Extract, Load, Transform) is incorrect because it loads raw data first and transforms it within the target system, providing less control over data quality and anonymization before warehouse loading.
   • DMPP (Massively Parallel Processing) is incorrect because it is an architecture for distributed query execution across multiple nodes, not a data pipeline pattern for moving data between systems.
3. Question 3Describe Core Data Concepts

   In an ELT process, raw data is loaded into the target system first and then transformed within that system.

   • ATrueCorrect
   • BFalse
   ✓ Correct answer: A

   ELT (Extract, Load, Transform) reverses the traditional ETL order-raw data is loaded directly into the target system first, and transformation happens within that system afterward. This approach has become popular with cloud data warehouses and big data platforms because they can handle transformation efficiently at scale, making the ELT process a valid alternative to traditional ETL.

   Why the other options are wrong

   • BFalse is incorrect. The statement is true. ELT (Extract, Load, Transform) reverses the traditional ETL order-raw data is loaded directly into the target system first, and transformation happens within that system afterward.
4. Question 4Identify Considerations for Relational Data

   Which Azure service provides a fully managed relational database service for open-source MySQL workloads?

   • AAzure Cosmos DB
   • BAzure SQL Managed Instance
   • CAzure Synapse Analytics
   • DAzure Database for MySQLCorrect
   ✓ Correct answer: D

   Azure Database for MySQL is a fully managed platform-as-a-service offering for the open-source MySQL relational database engine. Microsoft manages the underlying infrastructure, patching, backups, scaling, and high availability while you run standard MySQL workloads, making it the purpose-built service for hosting open-source MySQL databases in Azure.

   Why the other options are wrong

   • AAzure Cosmos DB is a NoSQL multi-model database, not a managed MySQL relational service.
   • BAzure SQL Managed Instance is a managed Microsoft SQL Server offering, not a MySQL service.
   • CAzure Synapse Analytics is a unified analytics/data-warehousing platform, not a managed open-source MySQL database.
5. Question 5Describe Considerations for Non-Relational Data

   What is the primary purpose of Azure Table Storage?

   • AStoring structured NoSQL key-attribute dataCorrect
   • BHosting virtual machine disk images
   • CStoring large binary objects like videos
   • DStoring relational data with complex joins
   ✓ Correct answer: A

   Azure Table Storage is designed for storing structured NoSQL data in a key-attribute (or key-value) format, where each entity is identified by a partition key and row key, and contains a collection of properties. It is optimized for storing large volumes of structured data that does not require complex queries or relationships, making it suitable for IoT sensor data, user profiles, application settings, and other semi-structured datasets. Table Storage is schema-less, allowing different entities to have different properties while maintaining a structured key-based organization.

   Why the other options are wrong

   • BHosting virtual machine disk images is incorrect because Azure Blob Storage (specifically page blobs) is the service designed for storing VHD files for virtual machines, not Table Storage.
   • CStoring large binary objects like videos is incorrect because Azure Blob Storage is the appropriate service for unstructured binary data like videos and large files, not Table Storage.
   • DStoring relational data with complex joins is incorrect because Table Storage is a NoSQL service without support for JOINs or relational schemas; Azure SQL Database is the appropriate service for relational data with complex queries.
6. Question 6Describe an Analytics WorkloadSelect all that apply

   Which TWO of the following are characteristics of a data warehouse? (Choose two.)

   • AOptimized for real-time transaction processing
   • BStores historical data for analytical queriesCorrect
   • CSes highly normalized schemas to reduce redundancy
   • DSupports complex aggregations and business intelligence queriesCorrect
   ✓ Correct answer: B, D

   A data warehouse is built for analytical workloads, so it is optimized to run complex aggregations and business intelligence queries over large volumes of integrated data. It consolidates and organizes data, often historically, to answer analytical questions efficiently. Supporting heavy aggregation and BI querying is a hallmark characteristic of a data warehouse.

   Why the other options are wrong

   • AOptimization for real-time transaction processing describes an OLTP system, which is the opposite of a warehouse's analytical focus.
   • CUsing highly normalized schemas to reduce redundancy characterizes OLTP design, whereas warehouses favor denormalized star or snowflake schemas.
7. Question 7Describe an Analytics WorkloadSelect all that apply

   Azure HDInsight provides managed clusters for popular open-source big-data frameworks. Select TWO frameworks you can run on Azure HDInsight.

   • AApache Kafka streaming ingestionCorrect
   • BMicrosoft Power BI reporting service
   • CApache HBase NoSQL data storeCorrect
   • DAzure Data Factory pipeline service
   ✓ Correct answer: A, C

   Azure HDInsight provisions managed clusters for open-source big-data frameworks, so you can run Apache Kafka for streaming ingestion and Apache HBase as a NoSQL wide-column store, along with Hadoop, Spark, and Hive. Microsoft manages the infrastructure while you use the standard open-source tools. Power BI and Azure Data Factory are Azure services, but they are not open-source frameworks that you deploy on an HDInsight cluster; they serve visualization and orchestration roles instead.

   Why the other options are wrong

   • BPower BI is a visualization and reporting service, not an open-source framework hosted on HDInsight.
   • DAzure Data Factory is a managed orchestration service, not an open-source framework you run on HDInsight.
8. Question 8Identify Considerations for Relational Data

   In a relational table, what is the primary purpose of a primary key?

   • ATo uniquely identify each row in the tableCorrect
   • BTo compress the data stored in the table
   • CTo grant users permission to read the table
   • DTo automatically encrypt every column in the table
   ✓ Correct answer: A

   A primary key is a column, or combination of columns, whose value uniquely identifies every row in a table. It enforces entity integrity: the values must be unique across all rows and cannot be null, which guarantees that each record can be reliably referenced. Primary keys are also commonly the target of foreign keys in other tables, establishing the relationships that make a relational model work.

   Why the other options are wrong

   • BCompression is handled by storage features such as data or page compression, not by the primary key.
   • CPermissions are managed through security objects and GRANT statements, not through a primary key.
   • DEncryption is provided by features like Transparent Data Encryption or Always Encrypted, and is not a function of a primary key.
9. Question 9Describe an Analytics Workload

   In a typical analytics workload, what is the main purpose of the data transformation step before data is loaded into a reporting model?

   • ATo clean, reshape, and standardize raw data so it is consistent and ready for analysisCorrect
   • BTo physically store the unaltered raw files in their original format
   • CTo render the final charts and dashboards shown to end users
   • DTo grant role-based access permissions to the published reports
   ✓ Correct answer: A

   Within an analytics pipeline, ingested raw data is often messy: it can contain inconsistent formats, duplicates, missing values, or fields that need to be combined or split. The transformation step (the 'T' in ETL/ELT) cleans and converts this data into a consistent, well-structured form that fits the target analytical model. This produces reliable, standardized data that reporting and visualization tools can query meaningfully, which is why visualization quality depends on proper transformation upstream.

   Why the other options are wrong

   • BStoring unaltered raw files is the ingestion/staging step, not transformation.
   • CRendering charts and dashboards is the visualization step that occurs after data is prepared.
   • DGranting access permissions is a governance/security concern, not the purpose of data transformation.
10. Question 10Identify Considerations for Relational Data

    Which statement accurately describes the difference between a clustered index and a nonclustered index in a relational table?

    • AA table has one clustered index (it sets physical row order) but many nonclustered indexes that point to the dataCorrect
    • BA table may hold many clustered indexes at once, yet it is strictly limited to a single nonclustered index across the whole table
    • CA clustered index keeps a full separate copy of every column of the table data, whereas a nonclustered index stores nothing at all
    • DA nonclustered index is the structure that fixes the physical storage order of the rows on disk within the underlying table
    ✓ Correct answer: A

    Because the clustered index determines how rows are physically sorted and stored, a table can have only one. Nonclustered indexes are separate structures that store key values and pointers to the underlying rows, and a table may have several of them to speed different queries.

    Why the other options are wrong

    • BThis inverts the limits: because a clustered index sets the physical row order, a table can have only one of it but many nonclustered indexes.
    • CA clustered index is the table's sorted data itself, not a separate copy, and a nonclustered index does store key values plus row pointers.
    • DThe clustered index, not the nonclustered index, determines physical row order; nonclustered indexes are separate structures pointing to the rows.

Who this DP-900 practice exam is for

This practice set is for anyone preparing for the DP-900: Azure Data Fundamentals exam at the foundational level - from first-time candidates building a foundation to experienced Microsoft practitioners doing a final review before test day. If you learn best by working through realistic questions and reading why each answer is right or wrong, it is built for you.

How to use this DP-900 practice exam

1. Start with the free sample questions above to gauge your current baseline.
2. Read the full explanation on every question, including why each wrong option is wrong.
3. Track your weak domains and focus your study where you are losing the most marks.
4. Once you are scoring consistently well, take a timed, full-length mock exam.
5. Use your readiness score to decide when you are ready to book the real DP-900 exam.

Related Microsoft resources

• DP-900 study guideKey concepts
• Microsoft practice examsAll Microsoft
• Certification pathWhere this fits
• Describe Core Data Concepts practice questions199 questions
• Identify Considerations for Relational Data practice questions177 questions
• Describe Considerations for Non-Relational Data practice questions137 questions
• Describe an Analytics Workload practice questions204 questions
• Certification exam guides & tipsBlog
• Plans & pricingFree & paid
• How these questions are written and reviewedMethodology
• Report a problem with a questionCorrections
• MB-310 practice examRelated
• MB-500 practice examRelated
• MB-800 practice examRelated

DP-900 practice exam FAQ