DP-900 Cheat Sheet
Azure Data Fundamentals — Quick reference for the exam
CLI Commands
27CREATE TABLE Customers (CustomerId INT PRIMARY KEY, Name VARCHAR(100) NOT NULL, City VARCHAR(50));TOPDDL: creates a table with a primary key and data types
ALTER TABLE Customers ADD Email VARCHAR(255);TOPDDL: changes the structure by adding a column
INSERT INTO Customers (CustomerId, Name, City) VALUES (1, 'Ana', 'Lima');DML: adds a row
UPDATE Products SET Price = Price * 1.10 WHERE Category = 'Beverages';TOPDML: changes existing rows. Without WHERE it would update every row
SELECT c.City, SUM(o.Amount) AS Total FROM Orders o JOIN Customers c ON o.CustomerId = c.CustomerId WHERE YEAR(o.OrderDate) = 2026 GROUP BY c.City ORDER BY Total DESC;TOPDML: JOIN between tables, filter with WHERE, aggregate with GROUP BY, sort with ORDER BY
GRANT SELECT ON Products TO analyst;TOPDCL: grants read permission (also DENY and REVOKE)
CREATE VIEW SalesByCity AS SELECT City, SUM(Amount) AS Total FROM Sales GROUP BY City;View: a virtual table based on a query
CREATE INDEX IX_Customers_LastName ON Customers (LastName);Index: speeds up lookups on that column
Comparisons
8| Type | Schema | Examples | Where in Azure |
|---|---|---|---|
| Structured | Fixed (rows and columns) | Customer and order tables | Azure SQL, PostgreSQL, MySQL |
| Semi-structured | Flexible | JSON, XML, YAML | Cosmos DB, Table storage |
| Unstructured | No schema | Images, video, audio, PDF | Blob Storage, ADLS Gen2 |
Quick Reference
29D1 · Core data concepts
Atomicity, consistency, isolation, and durability
DBA = operations, security, and backups. Engineer = pipelines. Analyst = models and reports
Key-value, document, column-family, and graph
Lake = raw files (schema-on-read). Warehouse = modeled data (schema-on-write)
Data lake + warehouse capabilities with Delta tables
Replicas may differ for a while, but they eventually match
D2 · Relational data on Azure
PK identifies each row. FK references another table's PK
Reduces redundancy and anomalies. 1NF = atomic values
CREATE, ALTER, DROP / SELECT, INSERT, UPDATE, DELETE / GRANT, DENY, REVOKE
vCore = independent compute and storage (serverless, Hyperscale). DTU = bundled measure
Reuses SQL Server licenses with Software Assurance (vCore model)
PostgreSQL and MySQL as Flexible Server. MariaDB retired on Sep 19, 2025
Transactional SQL database replicated automatically to OneLake
D3 · Non-relational data on Azure
Block (whole files), append (logs), page (VM disks)
Blob Storage + hierarchical namespace + POSIX ACLs
SMB/NFS file shares; Azure File Sync for on-premises caching
Key-value with PartitionKey + RowKey, no fixed schema
Normalized measure of the cost of each Cosmos DB operation
Strong, bounded staleness, session (default), consistent prefix, eventual
Current name of Azure Cosmos DB for MongoDB (vCore), since Nov 2025
D4 · Analytics
ETL transforms before loading. ELT loads raw and transforms in the destination
Single logical data lake per Fabric tenant, built on ADLS Gen2
Shortcuts = zero-copy access (ADLS, S3, GCS). Mirroring = continuous database replication
Data and AI platform on Apache Spark, with Unity Catalog
Still available; Microsoft's new investment goes to Fabric
Batch = accumulated, scheduled data. Streaming = each event as it arrives
Ingesting millions of events / stream processing with SQL
Power BI tables, relationships, DAX measures, and hierarchies (formerly "dataset")
Measure = computed at query time based on filters. Column = stored at refresh
Ready to practice?
Put what you learned to the test with original DP-900 exam questions.
Start DP-900 mock examUnlimited mock exams with the Pro plan