- The DP-900 Exam Snapshot
- Domain Weights at a Glance
- Domain 1: Core Data Concepts Cheat Sheet
- Domain 2: Relational Data on Azure
- Domain 3: Non-Relational Data on Azure
- Domain 4: Analytics Workload on Azure
- Logistics, Fees, and Retakes
- Confusable Pairs That Cost Points
- A Domain-Ordered Review Schedule
- Frequently Asked Questions
- DP-900 is Microsoft Certified: Azure Data Fundamentals; the passing mark is 700 or greater on a 1-1,000 scaled score, not 70% correct.
- Core data concepts and the analytics workload each carry the highest published range: 25-30%.
- Relational data on Azure is 20-25%; non-relational data is 15-20%.
- Standard retake rules: 24 hours after the first failure, 14 days after that, maximum five attempts per 12 months.
The DP-900 Exam Snapshot
This page condenses what Microsoft publishes about Microsoft Certified: Azure Data Fundamentals (DP-900) into one reviewable sheet. It is built around the current study guide ("Skills measured as of July 21, 2026"), so use it as a final-week refresher rather than a substitute for hands-on exploration of Azure data services. If you are earlier in your journey, start with the DP-900 study guide and come back here when you are ready to consolidate.
| Item | What Microsoft publishes |
|---|---|
| Certification | Microsoft Certified: Azure Data Fundamentals (exam DP-900) |
| Testing provider | Pearson VUE; Certiport is also offered to students and educators |
| Assessment time | 45 minutes for the assessment; 65 minutes is the Fundamentals seat time |
| Number of questions | Variable; no fixed DP-900 total or scored/unscored split is verified |
| Passing score | 700 or greater on a 1-1,000 scaled score |
| Format | Proctored, computer-based, potentially interactive; question-type mix is undisclosed |
| Prerequisites | No mandatory degree, work hours, or preceding certification; data and course familiarity recommended |
| Renewal | None; the Fundamentals certification does not expire |
Domain Weights at a Glance
The official skills outline lists four areas with percentage ranges. Use the ranges as published; do not convert them into exact question counts, because the live form's allocation is not disclosed. For a deeper breakdown, see the complete guide to all four DP-900 content areas.
| Domain | Published weight | Review priority |
|---|---|---|
| Describe core data concepts | 25-30% | Highest; foundation for everything else |
| Identify considerations for relational data on Azure | 20-25% | High; service selection and SQL basics |
| Describe considerations for working with non-relational data on Azure | 15-20% | Moderate; storage types and Cosmos DB |
| Describe an analytics workload on Azure | 25-30% | Highest; ingestion, warehousing, streaming, BI |
Two caveats from the guide itself: the bullets under each heading illustrate skills, and related topics may also appear. Also note a small wording discrepancy on Microsoft's own pages: the certification page and "Skills at a glance" call the fourth area "Describe an analytics workload on Azure," while the guide's detailed heading omits "on Azure." They describe the same area.
Domain 1: Core Data Concepts Cheat Sheet
This domain (25-30%) is vocabulary-heavy and rewards precision. Expect scenario wording that asks you to classify a dataset or pick the right kind of workload.
Describe core data concepts
Know the three shapes of data, the main data roles, and the difference between transactional and analytical processing.
- Structured data: fixed schema, rows and columns, typically in relational tables.
- Semi-structured data: self-describing but flexible, such as JSON, XML, or key-value documents.
- Unstructured data: no predefined model, such as images, video, audio, and free-form documents.
- Transactional (OLTP) workloads: many small reads and writes, strong consistency, row-oriented operations.
- Analytical (OLAP) workloads: large-scale aggregation over historical data, read-heavy, optimized for queries rather than updates.
- Common roles: database administrator (operations, security, backup), data engineer (pipelines and integration), data analyst (modeling and visualization).
Batch versus streaming
Batch processing collects data and processes it on a schedule or in large chunks; streaming processes records continuously as they arrive. Questions often describe a latency requirement ("alerts within seconds" versus "nightly report") and ask you to choose accordingly. Keep this contrast sharp; it returns in Domain 4.
File formats and storage ideas
Be able to recognize delimited text (CSV), JSON, XML, and columnar formats such as Parquet and ORC at a conceptual level. The key idea: row-oriented formats suit transactional access to whole records, while columnar formats suit analytical scans over a few columns across many rows.
Domain 2: Relational Data on Azure
At 20-25%, this domain tests whether you can describe relational concepts and match a requirement to the right Azure relational offering. It is less about writing complex queries and more about recognizing what each statement or object does.
Relational concepts to master
Tables, keys, normalization, and the common database objects.
- Tables, rows, columns: the basic structure; a primary key uniquely identifies a row, and a foreign key links to another table.
- Normalization: reduces redundancy by splitting data across related tables.
- Indexes: speed up lookups at the cost of extra storage and write overhead.
- Views: saved queries that present data as a virtual table.
- Stored procedures: reusable, server-side blocks of SQL logic.
- SQL statement categories: DDL defines structure (CREATE, ALTER, DROP); DML manipulates data (INSERT, UPDATE, DELETE, SELECT).
Choosing among Azure relational services
| Option | Think of it as | Typical fit |
|---|---|---|
| Azure SQL Database | Fully managed PaaS database | New cloud apps that want minimal administration |
| Azure SQL Managed Instance | PaaS with near-complete SQL Server compatibility | Migrating existing SQL Server workloads with minimal changes |
| SQL Server on Azure Virtual Machines | IaaS with full OS and instance control | Lift-and-shift where you need OS-level access |
| Azure Database for PostgreSQL / MySQL | Managed open-source relational engines | Apps built on those engines |
The decision logic the exam favors: the more management responsibility you want Azure to take, the further you move from VMs toward PaaS options. When a scenario stresses "least administrative effort," lean PaaS; when it stresses "full control" or "specific OS configuration," lean IaaS.
Domain 3: Non-Relational Data on Azure
The smallest domain (15-20%) still produces reliable points if you know which storage service matches which data shape.
Azure Storage and Cosmos DB essentials
Match the workload to the service, then to the model or API.
- Blob storage: unstructured objects such as images, video, backups, and logs; supports access tiers to balance cost and access frequency.
- Azure Data Lake Storage Gen2: built on blob storage with a hierarchical namespace, aimed at big-data analytics.
- Azure Files: managed file shares reachable over standard file-sharing protocols.
- Queue storage: simple message queuing between application components.
- Table storage: NoSQL key-value storage for structured, non-relational data.
- Azure Cosmos DB: globally distributed, multi-model NoSQL database with multiple APIs, including options for document, key-value, column-family, and graph data, plus compatibility APIs.
Non-relational design cues
Know why teams choose NoSQL: flexible schemas, horizontal scale, and low-latency access patterns. A scenario that mentions "varying document shapes," "global distribution," or "highly connected relationships" is pointing at Cosmos DB and its different data models. A scenario about "storing millions of images" points at blob storage, not a database.
Domain 4: Analytics Workload on Azure
This is tied for the heaviest range (25-30%) and the area where candidates with only database experience most often feel exposed. Learn the pipeline from ingestion to visualization.
Describe an analytics workload on Azure
Follow data from source to insight and name the service at each stage.
- Ingestion and orchestration: Azure Data Factory builds pipelines that move and transform data (ETL/ELT) from many sources.
- Large-scale analytics: Azure Synapse Analytics combines data warehousing and big-data analytics; Microsoft Fabric offers an integrated analytics platform; Azure Databricks provides an Apache Spark-based analytics environment.
- Data warehouse concepts: a warehouse stores integrated historical data modeled for reporting, often with fact tables (measurements) surrounding dimension tables (descriptive context).
- ETL versus ELT: transform before loading, or load first and transform inside the analytical engine.
- Real-time analytics: streaming sources such as Azure Event Hubs feed stream-processing services like Azure Stream Analytics for continuous queries.
- Visualization: Power BI connects to data sources to build reports and dashboards.
Batch pipeline versus streaming pipeline
| Aspect | Batch | Streaming |
|---|---|---|
| Latency | Minutes to hours or longer | Seconds or sub-second |
| Typical ingestion | Scheduled pipeline copies | Event ingestion service |
| Typical use | Nightly reporting, historical analysis | Telemetry, fraud signals, live dashboards |
Because Microsoft's product lineup evolves, rely on the current skills outline for exact service names and focus on the underlying concepts: ingest, store, process, model, visualize. If the wording of a service name looks unfamiliar, reason from the role it plays in the pipeline.
Logistics, Fees, and Retakes
These are the mechanics candidates ask about most, and the ones most often misreported on third-party sites.
- Fee: the issuer page's indexed rendering shows a U.S. list price of $99 USD. Country, tax, and eligible discounts affect what you actually pay, and Microsoft does not publish a general member/non-member tier. Details are in the DP-900 certification cost breakdown.
- Delivery: proctored computer-based exam through Pearson VUE, with online proctoring available in supported regions. Online delivery requires ID, system, and room checks. Request accommodations in advance.
- Resources: Microsoft Learn is not available during Fundamentals exams. Calculator availability and adaptive delivery were not verified, so do not plan around either.
- Scoring: 700 or greater on a 1-1,000 scale. Because it is scaled, it is not a simple 70% of questions, and some questions may be unscored. See the DP-900 passing score explainer.
- Retakes: 24 hours after a first failure, 14 days after any subsequent failure, with at most five attempts in 12 months. Payment applies unless an eligible offer covers it.
- Validity: the certification does not expire, and there is no renewal assessment or continuing-education quota.
- Prerequisites: none mandatory. Certiport's general pathway mentions 150 instructional hours, but that is not a verified DP-900 entry rule. Read the DP-900 requirements guide for the full picture.
Scheduling questions, including content updates by language and provider, are covered in the DP-900 exam dates guide. Note that provider-specific update dates listed by Certiport do not replace the Microsoft guide's July 21, 2026 skills-effective date.
Confusable Pairs That Cost Points
Fundamentals exams rarely ask for obscure trivia; they ask you to tell near-neighbors apart. Rehearse these contrasts until the answer is automatic.
- OLTP vs. OLAP: many small updates versus large analytical reads.
- DDL vs. DML: defining structure versus manipulating rows.
- Azure SQL Database vs. Managed Instance vs. SQL Server on a VM: managed single database, near-full SQL Server compatibility, or full OS control.
- Blob vs. Data Lake Storage Gen2: general object storage versus object storage with a hierarchical namespace for analytics.
- Cosmos DB vs. Table storage: a globally distributed multi-model database versus simple key-value table storage.
- ETL vs. ELT: transform before load versus after load.
- Batch vs. streaming: scheduled bulk processing versus continuous event processing.
- Fact vs. dimension tables: measurable events versus descriptive context.
Key Takeaway
Do not confuse this exam with AZ-900. DP-900 is the data-focused Azure Data Fundamentals exam; AZ-900 is Azure Fundamentals. If a practice resource spends most of its time on general cloud concepts like regions and pricing calculators, it is aimed at the wrong exam.
A Domain-Ordered Review Schedule
If you have about four weeks, ordering by dependency matters more than any particular study technique. Core concepts come first because Domains 2, 3, and 4 all reuse that vocabulary, and analytics comes last because it ties the earlier ideas together.
Core data concepts
- Memorize structured, semi-structured, and unstructured examples.
- Contrast OLTP and OLAP; batch and streaming.
- Learn the data roles and their responsibilities.
Relational on Azure
- Review keys, normalization, indexes, views, stored procedures.
- Practice DDL versus DML classification.
- Build a decision table for the Azure relational options.
Non-relational on Azure
- Map each storage service to its data shape.
- Review Cosmos DB models and why teams choose NoSQL.
Analytics workload and mixed review
- Trace a pipeline from ingestion to Power BI.
- Rehearse the confusable pairs above.
- Take timed mixed practice at the 45-minute pace on the DP-900 practice test site.
Practice banks are preparation material only; they cannot replicate the undisclosed live question mix or claim to contain official exam questions. For the broader plan, the first-attempt study guide goes deeper, and if you are weighing effort, the difficulty guide sets realistic expectations.
Frequently Asked Questions
You need 700 or greater on a 1-1,000 scaled score. This is not the same as answering 70% of questions correctly, and some questions may be unscored.
The number is variable, and no fixed DP-900 total has been verified. Microsoft's general 40-60 range is not an exact count for this exam, so prepare for variation rather than a set number.
Describe core data concepts and describe an analytics workload on Azure each carry the highest published range at 25-30%. Relational data is 20-25%, and non-relational data is 15-20%.
No. Microsoft Learn access is unavailable during Fundamentals exams. Plan to rely on what you have learned and practiced beforehand.
You may retake after 24 hours following the first failure and after 14 days for later attempts, up to five attempts per 12 months. Retake fees apply unless an eligible offer covers them.
Use this sheet as a checklist: if you can explain every bolded term above in a sentence and distinguish each confusable pair without hesitation, you are in a strong position. For career context after you pass, the ROI analysis and salary guide cover what the credential can and cannot do for your search.