The Microsoft Azure Data Fundamentals (DP-900) is Microsoft's entry-level certification for cloud data. It validates that you understand how data is stored, processed, and analyzed on Azure, covering relational databases, NoSQL services, and analytics platforms. No prior data engineering experience is required, making it accessible to developers, business analysts, students, and IT professionals who want a solid grounding in Azure data services before pursuing role-based certifications like DP-203 or DP-300.
This guide covers what the exam tests, which services to focus on, and how to study efficiently.
Exam Overview
| Detail | Value |
|---|---|
| Exam code | DP-900 |
| Questions | 40-60 |
| Time limit | 45 minutes |
| Passing score | 700 / 1000 (scaled) |
| Format | Multiple choice, drag-and-drop, build list, yes/no groups |
| Cost | $165 USD |
| Prerequisites | None |
| Certification validity | Lifetime (fundamentals certs don't expire) |
The 45-minute limit is comfortable for most candidates. The questions test conceptual recognition, not hands-on configuration. You won't need to write T-SQL or configure a Cosmos DB account from scratch. The focus is on knowing what each service does and which one fits a given scenario.
Exam Domains
| Domain | Weight |
|---|---|
| Describe core data concepts | 25-30% |
| Identify considerations for relational data on Azure | 20-25% |
| Describe considerations for working with non-relational data on Azure | 15-20% |
| Describe an analytics workload on Azure | 25-30% |
Core data concepts and analytics workloads each carry the most weight. Don't underestimate the analytics domain. Many candidates focus on databases and underestimate how much Synapse, Fabric, and Power BI content appears on the actual exam.
Core Services to Master
Core Data Concepts
This domain sets the vocabulary for the rest of the exam. You need to understand data structures well enough to classify them and recognize which storage approach fits each type.
- Structured data: Organized into rows and columns with a fixed schema. Relational database tables and CSV files with fixed headers are the classic examples.
- Semi-structured data: Has tags or markers but no rigid schema. JSON, XML, and YAML files fall into this category.
- Unstructured data: No predefined format. Images, audio files, and raw text documents are common examples.
- Transactional workloads (OLTP): Handle high volumes of short read/write operations. Think bank transactions or e-commerce orders. Optimized for consistency and speed on individual rows.
- Analytical workloads (OLAP): Handle large reads across historical data for reporting. Think dashboards and business intelligence. Optimized for aggregations, not individual row updates.
- Batch vs. streaming: Batch processing handles data at scheduled intervals. Streaming processes data in real time as it arrives.
Data roles also appear on the exam. A database administrator manages and optimizes databases. A data engineer builds and maintains pipelines. A data analyst interprets data and builds reports. Know what each role does and which Azure tools they typically use.
Relational Data on Azure
The exam focuses on Azure's relational database options and when to choose each one.
- Azure SQL Database: Fully managed PaaS database. Microsoft handles patching, backups, and scaling. The default choice for new applications on Azure that use SQL.
- Azure SQL Managed Instance: PaaS but with near-complete SQL Server compatibility. Good for lifting on-premises SQL Server workloads to the cloud without code changes.
- SQL Server on Azure VM: IaaS option where you manage the OS and SQL Server installation yourself. Maximum control, maximum responsibility.
- Azure Database for PostgreSQL / MySQL: Managed versions of popular open-source databases with Microsoft handling infrastructure.
Know your relational basics too: primary keys identify rows uniquely, foreign keys enforce relationships between tables, and indexes speed up reads at the cost of slightly slower writes. The exam may ask you to identify normalized data or spot a schema design pattern.
Non-Relational Data on Azure
Non-relational (NoSQL) databases trade rigid schemas for flexibility. Know the main options and their data models.
- Azure Cosmos DB: Globally distributed, multi-model database. Supports multiple APIs: Core SQL (JSON documents), MongoDB, Cassandra (column-family), Gremlin (graph), and Table (key-value). A single question might describe a graph workload or a MongoDB-compatible requirement and ask you to pick the right Cosmos DB API.
- Azure Blob Storage: Object storage for unstructured data like images, videos, and backups. Three access tiers: hot (frequent access), cool (infrequent access), archive (rare access, cheapest to store, most expensive to retrieve).
- Azure Table Storage: A simple key-value store inside Azure Storage accounts. Lower cost than Cosmos DB for basic lookups with no complex querying needs.
- Azure Files: Managed file shares accessible over SMB or NFS. Useful for migrating shared network drives to the cloud.
Analytics Workloads on Azure
This domain covers the modern data analytics stack on Azure. Focus on understanding the role each service plays rather than memorizing configuration details.
- Azure Synapse Analytics: Unified workspace combining a cloud data warehouse (dedicated SQL pools), serverless SQL queries over a data lake, and Apache Spark for big data processing. It's the hub for enterprise analytics on Azure.
- Microsoft Fabric: Microsoft's unified analytics platform that integrates data engineering, warehousing, real-time analytics, and Power BI into one SaaS service. Expect it to appear on the exam alongside Synapse.
- Azure Data Lake Storage Gen2 (ADLS Gen2): Hierarchical namespace built on top of Azure Blob Storage. Used to store raw and processed data for analytics pipelines.
- Azure Data Factory: Cloud-based ETL/ELT orchestration service. Moves and transforms data between sources and destinations on a schedule or trigger.
- Azure Databricks: Apache Spark platform optimized for collaborative notebooks and machine learning workloads.
- Power BI: Microsoft's visualization and reporting tool. Connects to Azure data sources, models relationships, and publishes dashboards and reports for business users.
Common Exam Traps
OLTP vs. OLAP mix-up: Questions often describe a workload and ask which storage type or service fits. If the scenario mentions high-volume transactions, individual row operations, or current operational data, that's OLTP. If it mentions aggregations, historical trends, or reporting and business intelligence, that's OLAP.
Cosmos DB API confusion: Cosmos DB supports multiple APIs and the exam picks scenarios requiring the right one. A social network with relationships between users points to the Gremlin (graph) API. A MongoDB-compatible workload points to the MongoDB API. A simple key-value lookup can use the Table API. Learn the data model associated with each API.
Azure SQL tier confusion: Azure SQL Database and Azure SQL Managed Instance are both PaaS, but Managed Instance offers near-full SQL Server compatibility for migration scenarios. SQL Server on Azure VM is IaaS. The exam tests whether you know which option fits a migration requirement vs. a greenfield deployment.
Blob storage access tiers: Archive tier is the cheapest to store data but costs more and takes longer to retrieve. Don't assume the cheapest storage tier means cheapest overall cost because retrieval costs are significant for archive.
ELT vs. ETL: ETL (extract, transform, load) transforms data before loading it into the destination. ELT (extract, load, transform) loads raw data first, then transforms it inside the destination system. Cloud data warehouses favor ELT because they have the compute to transform at scale.
Study Plan
| Week | Focus |
|---|---|
| 1 | Core data concepts: data types, OLTP vs. OLAP, batch vs. streaming, data roles |
| 2 | Relational data: Azure SQL options, normalization, indexes, views |
| 3 | Non-relational data: Cosmos DB APIs, Blob Storage tiers, Table Storage, Azure Files |
| 4 | Analytics workloads: Synapse, Fabric, ADLS Gen2, Power BI, Data Factory |
Most candidates with a general tech background can pass DP-900 in two to four weeks. If you're completely new to data concepts, spend an extra week on core concepts before moving to Azure services. Microsoft Learn covers all the material for free and is well-organized around exam objectives.
Recommended Resources
- Microsoft Learn: The official DP-900 learning path is free and maps directly to the exam domains.
- Official study guide: Microsoft publishes a DP-900 study guide listing every skill measured on the exam.
- Azure portal: Sign up for a free Azure account and browse Cosmos DB, Azure SQL, and Synapse in the portal. You don't need to configure anything; seeing the service structure helps exam questions make sense.
- Practice exams: Timed practice questions across all four domains are the most effective way to identify gaps before exam day.
Final Thoughts
DP-900 is a genuinely useful foundation, not just a checkbox cert. The concepts you learn here, from OLTP vs. OLAP to the difference between relational and non-relational data, come up in every data-related Azure certification that follows, whether that's DP-203 for data engineering, DP-300 for database administration, or PL-300 for Power BI.
Don't memorize service names in isolation. Practice mapping scenarios to services, and you'll be ready for the pattern-recognition style questions Microsoft uses on the exam.
Ready to test your knowledge? Try our DP-900 practice exams and see where you stand before exam day.