Free DP-900 practice exam: Azure Data Fundamentals exam guide (2026)
Format, the official domains in effect from July 2026, and the services that changed (Fabric, Flexible Server, Azure DocumentDB) so you can pass the DP-900.
The DP-900 (Azure Data Fundamentals) is the entry certification for data on Azure. It doesn't require prior experience, but it does expect you to understand the core concepts well: relational and non-relational data, transactional and analytical workloads, and which Azure service fits each case. It's the natural foundation for anyone aiming to become a data analyst, data engineer, or database administrator.
The current outline is from July 21, 2026. Since then Microsoft has retired some services and renamed others, so it's worth studying with up-to-date material.
What you need to know about the exam
| Questions | ~40-60 |
| Duration | 45 minutes |
| Reference cost | USD 99 (varies by country) |
| Passing score | 700 out of 1000 (scaled) |
| Languages | Offered in many languages, including Spanish |
| Validity | Does not expire |
| Prerequisites | None |
Official domains and their weight
| Domain | Weight | Main content |
|---|---|---|
| Core data concepts | 25-30% | Structured, semi-structured, and unstructured data, file formats, transactional and analytical workloads, data roles |
| Relational data on Azure | 20-25% | Normalization, SQL, database objects, the Azure SQL family, and open-source databases |
| Non-relational data on Azure | 15-20% | Blob Storage, Azure Files, Table storage, and Azure Cosmos DB |
| Analytics workloads on Azure | 25-30% | Ingestion and processing, analytical stores, Databricks, Microsoft Fabric, real time, and Power BI |
Domains 1 and 4 add up to more than half the exam. That's why it pays to master the concepts and the analytics part before memorizing services.
Domain 1: core data concepts (25-30%)
It's pure theory and shows up throughout the exam. You need to tell apart structured and unstructured data, the main data file formats (CSV, JSON, Parquet, Avro), OLTP and OLAP workloads, and the ACID properties. It also asks about roles: database administrator, data engineer, and data analyst.
Domain 2: relational data (20-25%)
It includes normalization, primary and foreign keys, and DDL, DML, and DCL SQL statements. For services, the Azure SQL family: Azure SQL Database, SQL Managed Instance, and SQL Server on virtual machines. For open source, Azure Database for PostgreSQL and Azure Database for MySQL.
Domain 3: non-relational data (15-20%)
It covers Azure Blob Storage and its access tiers, Azure Data Lake Storage Gen2, Azure Files, and Azure Table storage. The star is Azure Cosmos DB: its APIs, consistency levels, and Request Units.
Domain 4: analytics workloads (25-30%)
It asks about ETL and ELT, the data warehouse, the data lake, and the lakehouse, and about Microsoft Fabric (with OneLake) and Azure Databricks. For real time, the difference between batch and stream processing, Azure Stream Analytics, and Event Hubs. It closes with Power BI: the semantic model, the star schema, and which visual to choose.
Recent changes that can cost you points
- ▸Open source only with Flexible Server. The Single Server options for MySQL and PostgreSQL have retired, and Azure Database for MariaDB retired in September 2025.
- ▸Azure DocumentDB. The vCore option of Azure Cosmos DB for MongoDB is now called Azure DocumentDB.
- ▸Cosmos DB for PostgreSQL. Microsoft no longer recommends it for new projects; for distributed PostgreSQL it suggests Azure Database for PostgreSQL elastic clusters.
- ▸Microsoft Fabric gains weight. Synapse is still available, but Fabric is where Microsoft focuses investment, and the outline names it explicitly alongside Databricks.
- ▸"Dataset" is now "semantic model". The old Power BI name is no longer used.
Common mistakes that make people fail
- ▸Confusing a data lake with a data warehouse. The lake stores raw files; the warehouse stores modeled data in tables for analytical queries.
- ▸Picking the wrong Azure SQL option. New app → SQL Database; migrating SQL Server with minimal changes → Managed Instance; needing operating system access → SQL Server on a VM.
- ▸Mixing up the Cosmos DB APIs. Graphs → Gremlin; MongoDB apps → API for MongoDB; new app with JSON → NoSQL.
- ▸Forgetting the Blob tier minimums. Cool 30 days, Cold 90, and Archive 180, and Archive requires rehydration before reading.
How to prepare with Brains4Cloud
- ▸DP-900 cheat sheet: SQL, the Azure SQL family, Blob tiers, Cosmos DB APIs, Fabric, and Power BI visuals on a single page.
- ▸Data glossary: more than 35 DP-900 terms explained, with the context in which they appear on the exam.
- ▸Practice exam: 290 original questions aligned with the current outline, including single choice, multiple choice, fill-in, and matching, with an explanation for every option. Create your free account and take your first mock exam.
Frequently asked questions
Is the DP-900 hard? It's a fundamentals exam and doesn't require hands-on experience. The most demanding part is the variety of services and knowing which one fits each scenario. With two or three weeks of steady study it's very approachable.
Do I need to know SQL? Just the basics: recognizing statements such as SELECT, INSERT, UPDATE, or CREATE TABLE and knowing which category they belong to (DDL, DML, or DCL). You don't have to write complex queries.
Does the DP-900 expire? No. Microsoft fundamentals certifications don't expire or require renewal.
Which certification comes after the DP-900? It depends on your role: DP-700 (Fabric Data Engineer), DP-300 (database administration), or PL-300 (data analyst with Power BI). The DP-900 isn't a prerequisite for any of them, but it gives you the conceptual foundation.
Put what you learned to the test
Original practice exams with detailed explanations. No brain dumps.