How to Pass the Microsoft DP-900 Exam in 2026: A 4-Week Azure Data Fundamentals Study Plan for the New 21 July Objectives

The DP-900 exam is the fastest legitimate way to prove you understand cloud data, and it is also one of the easiest Microsoft certifications to underestimate. Forty-five minutes is not long, the questions move quickly between relational, non-relational and analytics territory, and Microsoft refreshed the skills measured on 21 July 2026. If you are searching for how to pass DP-900 without burning a month of evenings on the wrong material, this plan gives you four weeks, four domains and a clear order to work through them.

Here is the short answer. DP-900 is a beginner-level, 45-minute, proctored exam with roughly 40 to 60 questions, a pass mark of 700 out of 1000, and a list price of 99 US dollars. Most candidates who already work in IT need two to four weeks of part-time study. Complete beginners should take the full four weeks and not compress them. The plan below assumes six to eight hours a week, which is one weeknight session plus a longer weekend block.

What the DP-900 Exam Actually Looks Like in 2026

Before you plan anything, get the format straight. A lot of the advice floating around online describes an older version of the exam or confuses DP-900 with the retired DP-203 data engineering exam.

Two details matter more than people expect.

The first is the 45 minutes. That is genuinely tight if you are the kind of candidate who rereads every question twice. It works out at roughly 45 to 65 seconds per question. You do not have time to reason your way from first principles on more than a handful of items, which is why recall drilling matters more on DP-900 than on longer role-based exams.

The second is that DP-900 does not expire. Unlike the associate and expert Microsoft certifications, which need an annual free renewal assessment, fundamentals certifications are permanent once earned. That changes the cost calculation, because you pay once and keep the credential.

Exam Tip: Question types are not all straight multiple choice. Expect multi-select items, drag and drop matching, and short scenario questions. Use the free Microsoft exam sandbox before exam day so the interface is not new to you when the clock is running.

What Changed in the 21 July 2026 Objectives Update

Microsoft updated the DP-900 skills measured on 21 July 2026. This was a content refresh rather than a restructure, which is good news if you already started studying against the previous list.

The four functional groups and their weightings did not change. Three sub-areas were flagged as minor changes:

• Describe ways to represent data, inside the core data concepts domain
• Describe the capabilities of Azure storage, inside the non-relational domain
• Describe considerations for real-time data analytics, inside the analytics domain

The practical read is that Microsoft is keeping the exam aligned with the current Azure data estate, especially the continued shift of large-scale analytics towards Microsoft Fabric and Azure Databricks, both of which are named explicitly in the current objectives. If your study material still frames analytics primarily around Azure Synapse Analytics without mentioning Fabric, it is out of date.

Anything published before late July 2026 should be checked against the official skills measured list before you trust it. This is a common reason people fail: they study a syllabus that is eighteen months stale.

The Four DP-900 Domains and How to Weight Your Time

The exam is split into four domains with published weightings. Allocate your study hours in roughly the same proportions rather than spending equal time on each.

Notice that core data concepts and analytics are jointly worth up to 60 per cent of the exam. Candidates from a database background often over-invest in the relational domain because it feels familiar, then lose marks on analytics terminology they never touch at work. Do not make that trade.

The 4-Week DP-900 Study Plan

This plan assumes six to eight hours a week. Each week ends with active recall rather than more reading, because DP-900 rewards fast recognition of terminology.

Week 1: Core Data Concepts

This is the conceptual foundation, and it is the domain most likely to appear in disguised form inside other questions.

• Learn the three data shapes cold: structured (rows and columns, fixed schema), semi-structured (JSON, XML, Avro, Parquet, schema travels with the data) and unstructured (images, video, audio, free text documents).
• Learn the common file formats and what each is good for: CSV, JSON, XML, Avro, ORC and Parquet. Be able to say in one sentence why Parquet suits analytics (columnar, compressed) and why Avro suits row-based write-heavy ingestion.
• Nail the OLTP versus OLAP distinction. Transactional workloads are many small reads and writes with ACID guarantees. Analytical workloads are fewer, larger, read-heavy aggregations over historical data.
• Learn the three role definitions Microsoft uses. Database administrators manage availability, security, backup and performance. Data engineers build and manage pipelines and data stores. Data analysts model data and build visualisations.

Week 1 checkpoint: Without notes, write out a table mapping each of the three data shapes to two Azure services that would store it. If you cannot do it in five minutes, repeat the week's material before moving on.

Week 2: Relational Data on Azure

The relational domain rewards precision about which Azure product fits which scenario.

• Understand normalisation well enough to explain why it reduces duplication and what a normalised design looks like in practice. You will not be asked to normalise a complex schema, but you will be asked why you would.
• Recognise the core SQL statement families and what each does: SELECT, INSERT, UPDATE and DELETE for data manipulation, CREATE, ALTER and DROP for definition, and GRANT and REVOKE for control. Also know what a view, an index, a stored procedure and a foreign key are.
• Learn the Azure SQL family precisely, because this is a favourite source of scenario questions:
  • Azure SQL Database is fully managed platform as a service, best for new cloud applications with minimal administration.
  • Azure SQL Managed Instance gives near full SQL Server instance compatibility, best for lift and shift migrations that need instance-level features such as SQL Agent or cross-database queries.
  • SQL Server on Azure Virtual Machines is infrastructure as a service, where you keep full operating system control and full responsibility for patching.
• Know the open-source managed services: Azure Database for PostgreSQL and Azure Database for MySQL.

Week 2 checkpoint: Given five short migration scenarios, pick the right member of the Azure SQL family for each and justify it in one line. Scenario matching is exactly how this domain is examined.

Week 3: Non-Relational Data and Analytics Foundations

This week covers the smallest domain plus the start of the largest, so it is the heaviest reading week.

For the non-relational domain:

• Azure Blob storage for unstructured object data, with access tiers (hot, cool, cold and archive) that trade storage cost against retrieval cost and latency.
• Azure Files for fully managed SMB and NFS file shares that behave like a traditional network drive.
• Azure Table storage for cheap, schemaless key-value and semi-structured records.
• Azure Cosmos DB for globally distributed, low-latency NoSQL workloads. Know the API options (NoSQL, MongoDB, Cassandra, Gremlin and Table) and be able to pick the right one, for example Gremlin for graph relationships and Cassandra for column-family data.

For the analytics half:

• Learn the shape of an analytics pipeline: ingest, process, store, model, visualise.
• Distinguish a data warehouse (structured, schema on write, optimised for reporting), a data lake (raw, schema on read, cheap at scale) and a lakehouse (lake storage with warehouse-style structure and governance on top).
• Know Microsoft Fabric as the unified analytics platform with OneLake as its single storage layer, and Azure Databricks as the Apache Spark-based platform for large-scale processing and data science.

Week 3 checkpoint: Explain the difference between a data lake and a data warehouse to somebody non-technical in under a minute, then name one Azure service for each.

Week 4: Real-Time Analytics, Power BI and Exam Rehearsal

The final week is half new content, half rehearsal. Do not skip the rehearsal half.

• Learn batch versus streaming. Batch processes bounded data on a schedule with higher latency. Streaming processes unbounded data continuously with low latency. Know the real-time services by name, including Azure Stream Analytics, Azure Event Hubs and the real-time intelligence capability in Microsoft Fabric.
• Cover Power BI properly, because it carries real weight inside the analytics domain. Know what Power BI Desktop, the Power BI service and Power BI Mobile each do, what a dataset, report and dashboard are, and how a star schema of fact and dimension tables underpins a good model.
• Learn which visualisation suits which question. Bar and column charts compare categories, line charts show change over time, scatter plots show correlation between two measures, pie and donut charts show composition (and should be used sparingly), treemaps show hierarchical proportion, and key influencers surface drivers of a metric.
• Take the free Microsoft practice assessment and at least two timed full-length practice exams. Time them strictly at 45 minutes. If you are scoring consistently above 85 per cent on fresh questions, you are ready.

Exam Tip: Aim to finish the whole paper with five minutes to spare rather than aiming for perfection on every question. Flag anything that takes more than 90 seconds, move on, and return to the flagged items at the end. Marks lost to running out of time are the most avoidable failure mode on a 45-minute exam.

Five Topics That Catch DP-900 Candidates Out

These are the areas where otherwise well-prepared candidates lose marks.

1. Cosmos DB API selection. Knowing that Cosmos DB is a NoSQL database is not enough. You need to map a described workload (graph traversal, wide-column, document) to the correct API.
2. Blob storage access tiers. Archive tier is the cheapest to store and the most expensive and slowest to retrieve, with a rehydration delay measured in hours. Questions frequently hinge on that trade-off.
3. Managed Instance versus SQL Database. If a scenario mentions SQL Server Agent, cross-database queries, or minimal application change during migration, the answer is almost always Managed Instance.
4. Fabric versus Databricks versus Synapse. The current objectives name Fabric and Databricks explicitly. Older study material leans on Synapse. Learn the current framing.
5. Semi-structured data. Candidates routinely classify JSON as unstructured. It is semi-structured, because it carries its own schema markers even without a fixed table definition.

DP-900 or AZ-900: Which Should You Take First?

This comes up constantly, and there is no dependency between them. Neither is a prerequisite for the other.

If you have never touched Azure at all, AZ-900 first will make DP-900's service names land more easily. If you already work with data and simply need the cloud vocabulary, go straight to DP-900. If your target role is data engineering, DP-900 is the natural on-ramp to DP-700 Fabric Data Engineer, and if you are heading towards analytics and reporting it feeds directly into PL-300 Power BI Data Analyst.

One more piece of housekeeping. Microsoft has retired a long list of certifications through 2026, so check the full retirement list before you commit money to any Microsoft exam. DP-900 is not on it, which is part of why it remains a safe first credential.

Frequently Asked Questions

How difficult is the DP-900 exam?

DP-900 is a beginner-level exam and one of the more approachable Microsoft certifications, but it is not trivial. The difficulty comes from breadth rather than depth, since you need recognition-level knowledge of a wide spread of Azure data services in only 45 minutes. Candidates with any database or reporting experience typically pass with two to three weeks of part-time study.

What do you need to pass DP-900?

You need a scaled score of 700 or higher out of 1000. Microsoft does not publish how many raw questions that equates to, because the exam is scaled and questions are not equally weighted, so treat 700 as roughly a 70 to 75 per cent standard and aim comfortably above it in practice. There are no prerequisites and no experience requirement.

How long should I study for DP-900?

Four weeks at six to eight hours a week is a comfortable plan for a complete beginner. If you already work with SQL, data warehousing or Power BI, two weeks of focused revision is usually enough. The single biggest predictor of readiness is scoring consistently above 85 per cent on fresh, timed practice questions.

Are AZ-900 and DP-900 the same?

No. AZ-900 covers general Azure cloud concepts, core services, pricing and governance, while DP-900 covers data concepts and Azure data services specifically. Neither is a prerequisite for the other, and the overlap is limited to some shared cloud vocabulary.

Is DP-900 certification worth it?

For anyone entering a data role, yes. It costs 99 US dollars, never expires, takes weeks rather than months, and gives you the vocabulary that every Azure data conversation assumes you already have. It will not on its own land you a data engineering job, but it is a credible signal on a CV and a sensible stepping stone to DP-700 or PL-300.

Ready to Start Practising?

Reading the objectives will get you familiar with DP-900. Answering timed questions until the terminology is automatic is what gets you past 700 with minutes to spare.

CertCrush gives you exam-realistic practice questions with full explanations for every answer, so you learn why the wrong options are wrong rather than just memorising the right one. Work through the domains in the order set out above, drill each one until you are consistently above 85 per cent, then sit two full timed papers before you book.

Create your free CertCrush account and start practising today, or browse the full course catalogue to plan the certification path beyond DP-900.