Master the DP-900

An interactive study guide built on 7 memory techniques to help you pass the Microsoft Azure Data Fundamentals exam.

Core 25-30%Relational 20-25%Non-Relational 15-20%Analytics 25-30%

What it covers

Core data concepts (structured/semi/unstructured), relational databases (Azure SQL), non-relational (Cosmos DB), analytics workloads (Synapse, Power BI), and data governance (Purview).

Ideal for

IT professionals new to data, business analysts, project managers, and anyone wanting a foundational understanding of Azure data services.

Aspire to this if

You're starting a career in data, moving from infrastructure to analytics, or want to validate your Azure data knowledge before pursuing associate-level certifications.

Section 1 / Spatial Memory

The Map

Tap any component to see what you need to know for the exam.

🗄️ Relational Data

Azure SQL · PostgreSQL · MySQL

📄 Non-Relational Data

Cosmos DB · Blob · Table

📁 File & Streaming Data

Data Lake · Event Hubs

🏭 Azure Data Factory

ETL/ELT Orchestration

⚡ Azure Stream Analytics

Real-time Processing

🔥 Azure Databricks

Apache Spark Analytics

🗄️ Azure SQL Family

SQL DB · Managed Instance · VM

🌐 Azure Cosmos DB

Multi-model · Global Distribution

🏔️ Azure Data Lake

Storage Gen2 · Hierarchical

🔬 Azure Synapse Analytics

Unified Analytics · SQL + Spark + Pipelines

📊 Power BI

Visualisation & Dashboards

🔍 Microsoft Purview

Data Governance & Cataloguing

Section 2 / Narrative Memory

The Story

Follow the narrative to build a mental model of how everything connects.

🌊

The Data Ocean

Data exists everywhere — in databases, files, streams, and APIs. It comes in three forms: structured (tables with fixed schemas), semi-structured (JSON, XML with flexible tags), and unstructured (images, videos, documents). Understanding these forms is the first step.

Exam IntelThree data classifications: Structured (fixed schema, tables), Semi-structured (tags/keys but flexible — JSON, XML), Unstructured (no schema — images, video, audio). Know which Azure services handle each.

🏗️

The Foundations

Before data can flow, it needs a home. Relational databases store structured data in normalised tables with ACID guarantees. Non-relational databases offer flexibility — documents, key-value pairs, graphs, and column families each solve different problems.

Exam IntelACID = Atomicity (all or nothing), Consistency (valid state), Isolation (concurrent safety), Durability (committed = permanent). Normalisation reduces redundancy. Denormalisation improves read performance.

🗄️

The SQL Fortress

Azure SQL stands as a fortress for relational data. Three tiers serve different needs: SQL Database for cloud-native apps, Managed Instance for lift-and-shift migrations with near-full SQL Server compatibility, and SQL Server on VMs for those who need complete control.

Exam IntelAzure SQL Database = PaaS (Microsoft manages infra). Managed Instance = PaaS with 99% SQL Server compat. SQL VM = IaaS (you manage OS + SQL). Choose PaaS first, VM only when you need OS access or unsupported features.

🌐

The Global Network

Azure Cosmos DB spans the globe. Its multi-model engine speaks many languages — NoSQL, MongoDB, Cassandra, Gremlin, and Table. With single-digit millisecond reads and five tuneable consistency levels, it trades between performance and data freshness.

Exam IntelCosmos DB 5 consistency levels (strongest to weakest): Strong → Bounded Staleness → Session (default) → Consistent Prefix → Eventual. Request Units (RUs) = throughput currency. APIs determine data model.

🏭

The Pipeline Factory

Data rarely stays where it was born. Azure Data Factory builds pipelines that extract data from 90+ sources, transform it in mapping data flows, and load it into analytical stores. Triggers schedule the work — on a timer, in tumbling windows, or when events fire.

Exam IntelETL = Extract, Transform, Load (transform before storing). ELT = Extract, Load, Transform (load raw, transform in place — preferred for big data). Data Factory handles both. Mapping data flows = code-free visual transforms.

⚡

The Stream

Some data can't wait for batch processing. Event Hubs ingests millions of events per second, and Stream Analytics processes them in real-time using windowing functions — tumbling windows for fixed intervals, sliding windows for overlapping events.

Exam IntelBatch = process accumulated data at intervals. Streaming = process as it arrives. Windowing: Tumbling (fixed, non-overlapping), Hopping (fixed, overlapping), Sliding (event-triggered), Session (grouped by activity gap).

🏔️

The Data Lake

Azure Data Lake Storage Gen2 is the vast reservoir where raw data accumulates. Built on Blob Storage with a hierarchical namespace, it stores petabytes of data in formats like Parquet and Avro, ready for Spark and Synapse to analyse.

Exam IntelData Lake Gen2 = Blob Storage + hierarchical namespace. Blob tiers: Hot (frequent access), Cool (30+ days), Archive (180+ days, offline). Parquet = columnar (analytics). Avro = row-based (streaming). CSV/JSON = human-readable.

🔬

The Analytics Engine

Azure Synapse Analytics is the great unifier — SQL pools for data warehousing, Spark pools for big data processing, and built-in pipelines for orchestration. Serverless SQL lets you query Data Lake files without provisioning anything.

Exam IntelSynapse dedicated SQL pool = MPP data warehouse (provisioned, pay for compute). Serverless SQL pool = pay-per-query on files. Spark pool = big data notebooks. Synapse Link = HTAP (analytical queries on operational data without ETL).

📊

The Dashboard Gallery

Power BI transforms data into stories. Desktop for authoring, Service for sharing, Mobile for on-the-go consumption. Reports tell detailed multi-page stories; dashboards pin the most critical tiles on a single page for at-a-glance monitoring.

Exam IntelPower BI Dashboard vs Report: Dashboard = single page, pins from multiple reports, no filters. Report = multi-page, interactive filters, built from one dataset. Paginated reports = pixel-perfect, optimised for printing.

🔍

The Catalogue

Microsoft Purview brings order to the data estate. It scans sources automatically, builds a Data Map of all assets, traces lineage from source to dashboard, and applies sensitivity labels for compliance — ensuring data is discoverable, trustworthy, and governed.

Exam IntelPurview Data Catalog = discover/search data. Data Map = automated metadata scanning. Data Lineage = source-to-report tracing. Data Estate Insights = usage and health dashboards. Sensitivity labels from Microsoft 365 apply to data assets.

Section 3 / Acronym Memory

Mnemonic Wall

Memorable acronyms and phrases to anchor key exam concepts in your memory.

💎

ACID

Transaction guarantees. All-or-nothing, valid state, concurrent safety, committed = permanent.

📊

OLTP vs OLAP

OLTP = many small writes (point of sale). OLAP = complex reads (data warehouse). Know which services serve which.

🗄️

SMA

"SMV" — Azure SQL family from most managed to least. PaaS → PaaS → IaaS.

🌐

SBSCE

Cosmos DB consistency levels from strongest to weakest. Session is the default.

📄

DKCG

Four types of non-relational data stores. Cosmos DB supports all four via different APIs.

🔥

THSS

"THE windows HOP, SLIDE, and SESSION." Stream Analytics windowing functions.

🏔️

HCCA

Blob Storage access tiers. Hot = frequent. Cool = 30+ days. Cold = 90+ days. Archive = 180+ days, offline, must rehydrate.

📊

DRD

Power BI building blocks in order. Dashboards pin tiles from reports. Reports built from datasets.

🔑

PK-FK

PK uniquely identifies a row. FK references another table's PK. This creates relationships in relational databases.

📦

ELT

Modern approach: load raw data first, transform in place. Preferred for big data. ETL transforms before loading.

🔬

DSP

Synapse Analytics three compute options. Dedicated = provisioned MPP. Serverless = pay-per-query. Spark = big data.

📁

PJCA

Key file formats. Parquet = columnar (analytics). JSON = semi-structured. CSV = flat text. Avro = row-based (streaming).

🌐

NMCGT

Cosmos DB's 5 APIs. NoSQL is the native/recommended API.

🛡️

AAA

Three pillars of data security. Who are you, what can you do, what did you do.

🔍

CMLD

Microsoft Purview capabilities. Catalog (discover), Map (scan metadata), Lineage (trace flow), Labels (classify).

Section 4 / Contrast Memory

Versus Arena

Side-by-side comparisons to sharpen your understanding of similar concepts.

RelationalvsNon-Relational

Relational vs Non-Relational

Aspect	Relational	Non-Relational
Schema	Fixed (schema-on-write)	Flexible (schema-on-read)
Language	SQL	Varies (API-specific)
Scaling	Vertical (scale up)	Horizontal (scale out)
ACID	Full support	Varies by service
Best for	Structured data, transactions	Semi/unstructured, scale
Azure services	Azure SQL, PostgreSQL, MySQL	Cosmos DB, Table Storage

BatchvsStreaming

Batch vs Streaming

Aspect	Batch	Streaming
Timing	Processes accumulated data	Processes as it arrives
Latency	Minutes to hours	Seconds to milliseconds
Volume	Large datasets	Individual events
Examples	Nightly ETL, monthly reports	IoT sensors, fraud detection
Azure services	Data Factory, Synapse	Stream Analytics, Event Hubs
Complexity	Simpler to implement	More complex, stateful

SQL DBvsManaged InstancevsVM

SQL DB vs Managed Instance vs VM

Aspect	SQL DB	Managed Instance
Model	PaaS	PaaS \| IaaS
Compatibility	Most features	~99% SQL Server \| 100% SQL Server
Management	Fully managed	Mostly managed \| You manage
Use case	New cloud apps	Lift-and-shift \| Full control needed
OS access	No	No \| Yes
Cost	Lowest	Medium \| Highest

OLTPvsOLAP

OLTP vs OLAP

Aspect	OLTP	OLAP
Purpose	Day-to-day transactions	Analytics & reporting
Operations	Many small reads/writes	Complex queries, aggregations
Schema	Normalised (3NF)	Denormalised (star/snowflake)
Data	Current, operational	Historical, analytical
Examples	Order entry, banking	Data warehouse, BI
Azure services	Azure SQL, Cosmos DB	Synapse, Power BI

Star SchemavsSnowflake Schema

Star Schema vs Snowflake Schema

Aspect	Star Schema	Snowflake Schema
Dimensions	Denormalised	Normalised (sub-dimensions)
Joins	Fewer (faster queries)	More (complex queries)
Storage	More redundancy	Less redundancy
Query speed	Faster	Slower
Complexity	Simpler	More complex
Best for	Most BI/analytics	When storage matters

ETLvsELT

ETL vs ELT

Aspect	ETL	ELT
Transform where	In transit (staging)	In destination
Best for	Smaller datasets	Big data, cloud-native
Speed	Slower (transform first)	Faster load, transform later
Complexity	Transform engine needed	Leverage destination compute
Data quality	Clean before load	Raw data available
Azure example	Data Factory mappings	Synapse + Data Lake

DashboardvsReport

Dashboard vs Report

Aspect	Dashboard	Report
Pages	Single page	Multi-page
Source	Pins from multiple reports	Single dataset
Interactivity	Click to navigate	Filters, slicers, drill
Where created	Power BI Service only	Power BI Desktop or Service
Use case	At-a-glance monitoring	Detailed analysis
Refresh	Real-time tiles possible	Scheduled or on-demand

HotvsCoolvsColdvsArchive

Hot vs Cool vs Cold vs Archive

Aspect	Hot	Cool
Min retention	None \| 30 days	90 days \| 180 days
Storage cost	Highest \| Lower	Lower still \| Lowest
Access cost	Lowest \| Higher	Higher \| Highest (rehydrate)
Availability	99.9% \| 99%	99% \| Offline
Use case	Active data \| Short-term backup	Rarely accessed \| Compliance, long-term

Cosmos DB APIs Compared

Aspect	Cosmos DB APIs Compared
NoSQL	JSON documents	New apps (recommended)
MongoDB	BSON documents	MongoDB migrations
Cassandra	Column-family	Cassandra workloads
Gremlin	Graph (vertices/edges)	Relationship-heavy data
Table	Key-value	Azure Table migrations

Section 5 / Grouping Memory

Cheat Sheet

Organised reference grouped by exam domain — everything you need on one page.

Core Data Concepts

25-30%

Data Classification

Structured: fixed schema, tables, rows/columns
Semi-structured: flexible tags (JSON, XML, YAML)
Unstructured: no schema (images, video, audio, docs)

Data Roles

Database Admin: backups, access, performance, security
Data Engineer: pipelines, integration, data prep
Data Analyst: reports, dashboards, visualisations

File Formats

CSV: plain text, comma-separated, human-readable
JSON: key-value pairs, semi-structured
Parquet: columnar, compressed, analytics-optimised
Avro: row-based, schema evolution, streaming

Processing Patterns

Batch: scheduled, large volumes, high latency OK
Streaming: real-time, event-driven, low latency
ETL: transform before load (traditional)
ELT: load raw, transform in place (modern/cloud)

Relational Data

20-25%

Core Concepts

Tables, rows (records), columns (fields)
Primary Key: unique row identifier
Foreign Key: references another table's PK
Normalisation: reduce redundancy (1NF, 2NF, 3NF)
ACID transactions: all-or-nothing guarantees

SQL Commands

DDL: CREATE, ALTER, DROP (structure)
DML: SELECT, INSERT, UPDATE, DELETE (data)
DCL: GRANT, REVOKE (permissions)
Views: virtual tables from queries
Indexes: speed up queries, slow down writes

Azure SQL Family

SQL Database: PaaS, elastic pools, serverless
Managed Instance: near-full SQL Server compat
SQL VM: IaaS, full OS access, any SQL version
PostgreSQL/MySQL: open-source managed PaaS

Security

Azure AD authentication
Firewall rules (IP-based)
TDE: Transparent Data Encryption (at rest)
Always Encrypted: client-side encryption

Non-Relational Data

15-20%

Data Store Types

Key-value: simple lookups (Redis, Table Storage)
Document: JSON docs (Cosmos DB NoSQL)
Column-family: wide columns (Cosmos DB Cassandra)
Graph: vertices + edges (Cosmos DB Gremlin)

Azure Cosmos DB

Global distribution with multi-region writes
5 APIs: NoSQL, MongoDB, Cassandra, Gremlin, Table
5 consistency levels: Strong → Eventual
Request Units (RUs) = throughput measure
Partition key determines data distribution

Azure Storage

Blob: unstructured (Hot/Cool/Archive tiers)
File: SMB/NFS file shares (lift-and-shift)
Queue: messaging between components
Table: key-value NoSQL (simple, cheap)

When to Use

Key-value: session state, caching, config
Document: content management, catalogues
Column-family: IoT telemetry, time series
Graph: social networks, fraud detection

Analytics Workload

25-30%

Azure Synapse

Dedicated SQL pool: provisioned MPP warehouse
Serverless SQL pool: pay-per-query on files
Spark pool: big data notebooks
Synapse Link: HTAP (no-ETL analytics on ops data)

Power BI

Desktop: author reports locally
Service: publish, share, collaborate
Mobile: consume on devices
Dashboard = single page, pins from reports
Report = multi-page, interactive, one dataset
Paginated = pixel-perfect, print-ready

Data Warehousing

Star schema: fact table + dimension tables
Snowflake schema: normalised dimensions
Fact tables: measurements/events (numeric)
Dimension tables: descriptive attributes
Slowly Changing Dimensions (SCD) types

Governance

Microsoft Purview: unified data governance
Data Catalog: discover and classify assets
Data Map: automated metadata scanning
Data Lineage: trace data flow end-to-end
Sensitivity labels for compliance

Section 6 / Method of Loci

The Memory Palace

Walk through themed rooms — each object anchors a concept in spatial memory.

The Foundation Hall

Core Data Concepts — Where understanding begins

📊

Structured Data

Tables with rows and columns. Fixed schema. SQL databases

📄

Semi-Structured

JSON, XML, YAML. Flexible tags/keys. No fixed table format

🖼️

Unstructured Data

Images, video, audio, documents. No inherent schema

💎

ACID Properties

Atomicity (all/nothing), Consistency, Isolation, Durability

📦

ETL vs ELT

Transform then load (ETL) vs load then transform (ELT). ELT preferred for cloud

⚡

Batch vs Stream

Batch = scheduled bulk. Stream = real-time. Different tools for each

The SQL Chamber

Relational Data — Where structure reigns

🗄️

Azure SQL Database

Fully managed PaaS. Elastic pools for multi-tenant. Serverless option

🔄

Managed Instance

Near 100% SQL Server compat. Lift-and-shift migrations. VNet integration

🖥️

SQL VM

Full IaaS control. Any SQL Server version. OS-level access

🔑

Keys & Relations

Primary Key = unique ID. Foreign Key = reference. Creates table relationships

📐

Normalisation

1NF → 2NF → 3NF. Reduces redundancy. Improves data integrity

🐘

Open Source DBs

Azure Database for PostgreSQL and MySQL. Fully managed PaaS

The Cosmos Chamber

Non-Relational Data — Where flexibility thrives

🌐

Cosmos DB

Global distribution. <10ms latency. 5 APIs. 5 consistency levels

📄

Document Store

JSON documents. Flexible schema. Each doc can differ. Cosmos DB NoSQL API

🔑

Key-Value Store

Simplest model. Fast lookups by key. Azure Table Storage, Redis Cache

📊

Column-Family

Wide columns. Good for time-series, IoT. Cosmos DB Cassandra API

🕸️

Graph Store

Vertices + edges. Model relationships. Cosmos DB Gremlin API

📦

Blob Storage

Unstructured data. Hot/Cool/Archive tiers. Block, append, page blobs

The Pipeline Room

Data Integration — Where data flows

🏭

Data Factory

90+ connectors. Pipelines. Mapping data flows. Schedule/event triggers

⚡

Stream Analytics

SQL-like queries on streams. 4 windowing functions. Real-time output

🔥

Azure Databricks

Apache Spark platform. Notebooks. Python/Scala/SQL/R. Delta Lake

🔬

Synapse Analytics

Unified: SQL pools + Spark pools + Pipelines. Serverless & dedicated

🏔️

Data Lake Gen2

Blob + hierarchical namespace. Parquet/Avro/Delta. Big data optimised

📡

Event Hubs

Millions of events/second. Kafka compatible. Streaming ingestion

The Insight Gallery

Analytics & Governance — Where value emerges

📊

Power BI Desktop

Author reports. Connect to data. DAX calculations. Publish to Service

📋

Power BI Service

Share reports. Create dashboards. Schedule refresh. Collaborate

⭐

Star Schema

Fact table (measures) + dimension tables (attributes). Analytics standard

🔍

Microsoft Purview

Data Catalog + Map + Lineage. Sensitivity labels. Unified governance

🛡️

Data Security

Azure AD auth. Firewall rules. TDE at rest. Always Encrypted

📈

Paginated Reports

Pixel-perfect. Print-optimised. Multi-page tables. SSRS-based

Section 7 / Pattern Recognition

Pattern Spotter

Decision trees and trigger-answer pairs — see the pattern, know the answer.

Which Azure Database Service?

Which Azure Database Service?
  ├── Relational + new cloud app → Azure SQL Database
  ├── Relational + migrating SQL Server → SQL Managed Instance
  ├── Relational + need OS access → SQL Server on VM
  ├── NoSQL documents + global scale → Cosmos DB
  ├── Simple key-value lookups → Azure Table Storage
  ├── PostgreSQL or MySQL → Azure Database for PostgreSQL/MySQL
  └── Unstructured files/blobs → Azure Blob Storage

Which Analytics Service?

Which Analytics Service?
  ├── Full data warehouse (provisioned) → Synapse Dedicated SQL Pool
  ├── Ad-hoc queries on Data Lake files → Synapse Serverless SQL Pool
  ├── Big data processing with Spark → Synapse Spark Pool / Databricks
  ├── Real-time stream processing → Azure Stream Analytics
  ├── Business intelligence dashboards → Power BI
  └── Data governance & cataloguing → Microsoft Purview

Which Cosmos DB API?

Which Cosmos DB API?
  ├── New app, JSON documents → NoSQL API (recommended)
  ├── Migrating from MongoDB → MongoDB API
  ├── Wide-column / Cassandra workload → Cassandra API
  ├── Graph / relationship-heavy data → Gremlin API
  └── Migrating from Azure Table Storage → Table API

Which File Format?

Which File Format?
  ├── Analytics / columnar queries → Parquet
  ├── Streaming / schema evolution → Avro
  ├── API / semi-structured exchange → JSON
  ├── Simple tabular / human-readable → CSV
  └── Delta Lake / ACID on data lake → Delta (Parquet + transaction log)

Decision Cards

“globally distributed” or “multi-region writes”→Azure Cosmos DB

“single-digit millisecond latency”→Cosmos DB

“Request Units” or “RU/s”→Cosmos DB throughput

“ACID transactions” on relational data→Azure SQL Database

“lift-and-shift” SQL Server migration→SQL Managed Instance

“elastic pool” for multi-tenant→Azure SQL Database

“hierarchical namespace”→Data Lake Storage Gen2

“Hot, Cool, Archive” tiers→Azure Blob Storage

“tumbling window” or “hopping window”→Stream Analytics windowing

“mapping data flow” or “90+ connectors”→Azure Data Factory

“serverless SQL pool” or “pay-per-query”→Synapse Analytics

“Synapse Link” or “HTAP”→No-ETL analytics on operational data

“star schema” or “fact and dimension”→Data Warehouse design

“Data Catalog” or “data lineage”→Microsoft Purview

“dashboard vs report”→Dashboard = single page, Report = multi-page

“paginated report” or “pixel-perfect”→Power BI Paginated Reports

“partition key” for NoSQL→Cosmos DB data distribution

“graph database” or “vertices and edges”→Cosmos DB Gremlin API

Ready to certify?

Train with practitioners, not presenters

Lucid Labs delivers Microsoft certification training led by Microsoft Certified Trainers (MCTs) and grounded in real-world project experience. We adapt every session to your team's environment, data stack, and business objectives — because the best exam prep comes from engineers who build these solutions every day.

🎯

Tailored Content

Training built around your actual data, your tools, and your use cases — not generic slides.

🛠️

Hands-On Labs

Work through real scenarios in your own environment with expert guidance at every step.

📈

Exam + Capability

Pass the exam and build lasting skills your team can apply from day one.

Talk to us about Azure Data Fundamentals training

Custom training for teams & individuals — remote or on-site across Australia

Keith Oak

Chief Technology Officer — Lucid Labs

Microsoft Solutions Partner architect specialising in Fabric, Azure Data & AI, and GitHub Enterprise. 18+ years delivering data platforms for Australian businesses — building the systems these exams test every day.

LinkedIn ↗lucidlabs.com.au ↗Published 29-03-2026