A practical route through Microsoft Azure Data Fundamentals: what the exam actually tests, the four skills domains and their weights, a realistic four-week study plan, the topics that trip people up, and where to go once you have it.
Updated July 2026·~8 min read
The exam
What is DP-900, and is it for you?
DP-900 earns the Microsoft Certified: Azure Data Fundamentals credential — the entry-level exam for anyone working near data in Azure. It's aimed at candidates with basic familiarity with relational versus non-relational data and transactional versus analytical workloads, and it does not assume database-administration experience. Common candidates include data analysts moving into Power BI work, software engineers crossing into a data role, and product or operations partners who need to discuss data architecture credibly.
The exam does not ask you to write a SQL query from scratch or design a schema. It expects you to recognise scenarios: structured versus semi-structured versus unstructured data, OLTP versus OLAP workloads, when a lift-and-shift SQL Server on an Azure VM beats Azure SQL Database, when to reach for Azure Cosmos DB instead of Azure Blob Storage, and how Microsoft Fabric's analytics surface relates to Azure Synapse Analytics.
Microsoft last updated the DP-900 skills outline on 1 November 2024, and it is the recommended, though optional, on-ramp to Microsoft's Data and BI role-based tracks. See the full breakdown, including every question type the app supports, on the DP-900 study app page, or read the official outline at learn.microsoft.com.
Four domains, weighted from Microsoft's November 2024 outline. Two of them — core concepts and analytics — are tied for the largest share, so plan your time accordingly rather than splitting study evenly across all four.
DP-900 skills measured (November 2024 outline)
Domain
Weight
Describe core data concepts
25–30%
Identify considerations for relational data on Azure
20–25%
Describe considerations for non-relational data on Azure
15–20%
Describe an analytics workload on Azure
25–30%
Core data concepts is the conceptual base: structured versus semi-structured versus unstructured data, common file formats such as CSV, JSON, Parquet, and Avro, database types from relational to graph, and OLTP versus OLAP. Analytics workloads is the applied end: Microsoft Fabric and Azure Synapse Analytics, ingestion with Azure Data Factory, streaming with Azure Stream Analytics, and Power BI. Together they're over half the exam, so treat the two smaller domains — relational and non-relational data — as focused sprints rather than the main event.
Four-week study plan
A realistic four-week study plan
Most candidates with some IT background pass DP-900 after one to two weeks of focused study. If data and cloud concepts are genuinely new to you, stretch that to three or four weeks — the plan below assumes the longer end and maps one week onto each domain, finishing with simulation.
Core data concepts
The largest domain and the conceptual base everything else assumes. Structured, semi-structured, and unstructured data; file formats including CSV, JSON, Parquet, and Avro; database types from relational and key-value to document, graph, and columnar; the OLTP-versus-OLAP split; and the four classic data roles — administrator, engineer, analyst, and scientist.
Relational data on Azure
Relational concepts such as tables, indexes, views, and normalisation, then the SQL Server family on Azure: Azure SQL Database, Azure SQL Managed Instance, and SQL Server on Azure VMs, plus the open-source options for PostgreSQL and MySQL. Drill scenario questions like "single-tenant lift-and-shift" against "cloud-native serverless app" until the right service is automatic.
Non-relational data on Azure
The smallest domain, so it should take the least time. Azure Cosmos DB and its API options — NoSQL, MongoDB, Cassandra, Gremlin, and Table — then Azure Storage services for non-relational data: Blob containers, file shares, Azure Data Lake Storage Gen2, and table storage. Focus on matching an access pattern to the right storage tier.
Analytics workloads and simulation
Tied for the largest domain. Modern data-warehousing concepts; Microsoft Fabric and Azure Synapse Analytics; ingestion with Azure Data Factory and Synapse pipelines; real-time streaming with Azure Stream Analytics; and Power BI, including the Power BI Desktop versus Power BI Service distinction. Close the week with two full-length, timed simulator runs, and book the exam once your readiness sits comfortably above the pass mark, not just at it.
Where people lose marks
The hardest DP-900 topics
DP-900 is a Fundamentals exam, but a handful of areas still cost candidates marks — usually because two or three services sound similar and the exam expects you to pick the right one for a scenario, not just recognise the name.
Relational versus non-relational trade-offs
The exam frequently sets up a scenario and expects you to decide whether it belongs in a relational store (Azure SQL Database, Managed Instance, SQL Server on Azure VMs) or a non-relational one (Azure Cosmos DB, Azure Storage). The trap is that both can technically hold the data; the question is testing whether you understand schema rigidity, query patterns, and scale requirements, not whether the data could ever fit either way.
The Azure SQL service triangle
Azure SQL Database, Azure SQL Managed Instance, and SQL Server on Azure VMs solve overlapping problems at different levels of control. A cloud-native app with minimal management overhead points to Azure SQL Database; a near-verbatim lift-and-shift of an on-premises instance with cross-database queries points to Managed Instance; and a need for OS-level access or unsupported SQL Server features points to the VM option. Learn the deciding factor, not just the three names.
Cosmos DB's API options
Azure Cosmos DB is one service with five API surfaces — NoSQL, MongoDB, Cassandra, Gremlin, and Table — and DP-900 likes to describe a workload (document storage, wide-column, graph traversal) and ask which API fits. Memorise the pairing of workload shape to API name; the underlying service is the same, so the question is really about data-model recognition.
Batch versus streaming ingestion
Azure Data Factory and Synapse pipelines move data on a schedule or in response to an event — batch ingestion. Azure Stream Analytics processes data continuously as it arrives — streaming ingestion. Questions about "near real-time dashboards" or "nightly reporting" are testing whether you reach for the right ingestion pattern, not just whether you can name both services.
The analytics service landscape
Microsoft Fabric, Azure Synapse Analytics, and Power BI overlap in places, and the naming does candidates no favours. Fabric is Microsoft's unified analytics platform built around OneLake; Synapse Analytics is the underlying big-data and data-warehousing engine; Power BI is the visualisation layer, with Power BI Desktop for authoring reports and Power BI Service for publishing and sharing them. Keep the four names straight and DP-900's analytics domain gets much easier.
How to practise
Practise the way the exam works
Courses and videos leave you able to spot a correct answer, not summon one, yet DP-900's scenario questions want the summoning done under mild time pressure. Two habits close that gap. Start with retrieval practice: give your answer from memory before checking the explanation, because pulling the fact back is what fixes it — doubly true on a vocabulary-heavy paper like DP-900, where quick recognition is half the battle. Pair it with spaced repetition: return to each domain just as it begins to fade rather than covering it once and moving on. Four domains is easy to hold for a day and harder to hold across four weeks.
Then close with full-length simulation. A timed, mixed-domain run at DP-900's actual 45-minute length is the only way to find out whether you can keep pace across 40 to 60 questions. Do at least two before you book, and review every wrong answer properly rather than just noting the score. Case-study-style linked questions do appear on DP-900, mostly around storage and identity, so make sure one simulator run includes them.
Azure Mastery rebuilds its study plan from your answer history, surfacing the weakest domains first.
Build all three habits in one app
Azure Mastery is built around exactly this loop for DP-900. Practise with over 350 exam-style questions mapped to the four November 2024 domains, get an on-device readiness score from the Exam IQ engine, follow an adaptive study plan that targets your weak spots, and let knowledge-decay alerts tell you what to revise before you forget it. For a proper rehearsal, the Practice tab hosts a timed paper built to the exam's true length and pacing, and the whole app works offline with no account needed to begin.
DP-900 does not expire and it is not a formal prerequisite for anything, so think of it as the on-ramp that confirms data is the right direction before you commit two or three months to a role-based exam. Two paths lead on from here, and the split is administration versus engineering.
If you want to run and secure Azure SQL workloads in production — configuring high availability and disaster recovery, tuning performance, managing security — the next stop is DP-300 (Azure Database Administrator Associate). It builds directly on the relational-data domain here, but expects hands-on T-SQL and configuration rather than scenario recognition.
If you want to build the pipelines that load and shape data, you have a choice of platform. DP-700 (Fabric Data Engineer Associate) goes deep on Microsoft Fabric: Dataflows Gen2, notebooks and PySpark, streaming with Eventstream, and Lakehouse and Warehouse optimisation. DP-750 (Azure Databricks Data Engineer Associate) covers the same ground on Azure Databricks instead — Lakeflow Connect, Unity Catalog governance, and Spark job tuning. Most engineers pick one platform to match their employer's stack; some eventually sit both.
Not sure administration or engineering is the right fit, or whether Azure's general infrastructure track suits you better first? The Which Microsoft certification first? guide walks through the Azure, security, and data tracks side by side.
Frequently asked
DP-900 FAQs
How hard is the DP-900 exam?
DP-900 is one of the more approachable Microsoft Fundamentals exams because it stays conceptual rather than hands-on. It expects you to recognise scenarios — structured versus semi-structured data, OLTP versus OLAP, which Azure SQL service fits a migration, which storage tier fits an access pattern — rather than write SQL or configure a live resource. Most candidates with some IT background pass after one to two weeks of focused study; if data concepts are genuinely new, three to four weeks is more realistic.
How long should I study for DP-900?
Plan for one to two weeks of focused study if you already work in IT or software. If data and cloud concepts are new to you, allow three to four weeks, and lean on the four-week plan on this page. The exam rewards recognising the right service for a scenario over memorising definitions, so spend your time on comparisons — Azure SQL Database versus Managed Instance, Cosmos DB API choice, Microsoft Fabric versus Azure Synapse Analytics — rather than flashcards.
Do I need DP-900 before DP-300, DP-700, or DP-750?
You need 700 out of 1000 to pass. The score is scaled, not a raw percentage, so it does not translate directly to a fixed number of correct answers out of the roughly 40 to 60 questions. Treat 700 as a comfortable target rather than a line to scrape over, since a handful of unfamiliar questions will not sink an otherwise solid attempt.
Does the DP-900 certification expire?
No. Microsoft Fundamentals certifications, including AZ-900, DP-900, AI-901, MS-900, and PL-900, do not expire once you pass. This is different from Associate and Expert certifications such as DP-300 or DP-700, which require a free renewal assessment on Microsoft Learn every year.
How much does DP-900 cost and can I retake it if I fail?
Exam pricing varies by country, so check Microsoft's exam page for the fee where you live rather than trusting a figure you read elsewhere. If the result does not go your way, you can re-sit after 24 hours, then wait 14 days before each of the two attempts after that, with Microsoft's ceiling set at five sittings in any twelve-month window. You book and pay for each attempt separately.
Keep reading
Related guides
Deciding whether DP-900 is the right first exam, or where a sibling Fundamentals credential fits? These guides help you choose, and the DP-900 app page has the full breakdown of what is inside Azure Mastery.
Practise with over 350 exam-style DP-900 questions across all four domains, with an on-device readiness score, an adaptive study plan, and a full-length simulator. Free to start, works offline.
