How to Pass DP-900: The 2-Week Study Plan

DP-900 is a genuinely beginner-friendly exam — two weeks of focused study is realistic for most people, less if you already work with data. The trap is the breadth of the analytics domain: Synapse, Fabric, Databricks, Event Hubs, Stream Analytics, and Power BI all look similar until you learn the one-line job description of each.

Week 1 — core concepts + relational/non-relational

• Day 1–2: data types (structured/semi/unstructured), OLTP vs OLAP, batch vs streaming, ETL vs ELT — the vocabulary foundation
• Day 3–4: relational model — keys, normalization, views, indexes, basic SQL statement roles; Azure relational services and when each fits (SQL Database vs Managed Instance vs VM vs PostgreSQL/MySQL)
• Day 5–6: non-relational — Cosmos DB (APIs, consistency levels, partition keys), Blob tiers, ADLS Gen2 vs Blob, Table Storage vs Cosmos Table API, Azure Files
• Day 7: review + first pass through our question bank — target the core-concepts and relational topics

Week 2 — analytics + consolidation

• Day 8–9: the analytics stack — warehouse vs lake vs lakehouse, Synapse components, Fabric, Databricks, Data Factory
• Day 10: real-time — Event Hubs vs IoT Hub vs Stream Analytics; windowing basics; medallion architecture (bronze/silver/gold)
• Day 11: Power BI + fact/dimension/star schema concepts
• Day 12–13: full question-bank pass + mock exam; drill every wrong answer
• Day 14: cheat-sheet skim, Cosmos consistency levels one more time, exam

Coming from AWS? Translation table

• RDS / Aurora → Azure SQL Database (+ Managed Instance for near-full compat)
• DynamoDB → Cosmos DB (note Cosmos is multi-model — NoSQL/Mongo/Cassandra/Table/Gremlin APIs)
• S3 → Blob Storage / ADLS Gen2; S3 tiers → Hot/Cool/Cold/Archive
• Redshift → Synapse dedicated SQL pools (or Fabric warehouse)
• Kinesis → Event Hubs; Kinesis Data Analytics → Stream Analytics
• Glue → Azure Data Factory; QuickSight → Power BI
• EMR → Azure Databricks / HDInsight / Synapse Spark

What 'ready' looks like

You're ready when service-selection is reflex: 'JSON documents at global scale' → Cosmos DB; 'SMB file share in the cloud' → Azure Files; 'event ingestion millions/sec' → Event Hubs; 'interactive dashboards' → Power BI. And you can recite the five Cosmos consistency levels in order — that question shows up for almost everyone.