HomeCloud FoundationsAzure Architecture
Azure Architecture Guide

Azure Architecture: trade-offs, data, AI and cloud patterns

7 decks to go beyond Azure fundamentals. Cloud patterns, modern data architectures, industrialized machine learning and event-driven architectures — the topics of an Azure data architect. This guide approaches these topics from the angle of choices and implementation specific to Azure.

Prerequisite: Azure Fundamentals level or equivalent cloud experience recommended.
7decks
343flashcards
Advancedlevel
Post-Fundamentalsprerequisite

What You Will Understand

This guide covers the major themes of Azure data architecture. It is not aligned with a single official certification but with the skills expected of a senior data architect or data engineer on the Azure platform. For a vendor-neutral explanation of the underlying concepts (Data Lake, Data Mesh, governance, MLOps), see the "Further Reading" section below.

Cloud architecture patterns

Lambda, Kappa, Medallion, microservices, event-driven — key design patterns applicable on Azure and across cloud ecosystems.

Modern data architectures

Data Lake, Lakehouse, Data Mesh, Data Products. How these data organization paradigms translate concretely into Azure services.

Azure analytics platforms

Positioning and comparing Databricks, Azure Synapse Analytics and Microsoft Fabric: when to choose what, and why.

Data governance and management

Microsoft Purview, Data Catalog, data lineage, classification and governance policies in an Azure architecture.

Streaming, IoT and event-driven architectures

Event Hubs, Service Bus, Stream Analytics, Azure IoT Hub — real-time and event-driven architectures on Azure.

AI and MLOps on Azure

Azure Machine Learning, MLflow, Responsible AI, model industrialization and the ML lifecycle in production.

The 7 Guide Decks

Each deck addresses a key domain of Azure data architecture. The suggested order moves from cross-cutting concepts to the most specialized topics.

1
Cloud Architecture Patterns

Key cloud and data design patterns: Lambda, Kappa, Medallion, CQRS, Event Sourcing, Saga, Strangler Fig. Architectural references applicable regardless of the cloud platform, illustrated with Azure examples.

50 cards
2
Data Lake & Lakehouse

Classic Data Lake vs modern Lakehouse architectures on Azure. ADLS Gen2, Delta Lake, file formats (Parquet, Avro, ORC), partitioning and read optimization strategies.

45 cards
3
Databricks / Synapse / Fabric

Comparing and positioning the three major Azure analytics platforms. When to use Databricks (ML, complex data engineering), Synapse (DWH, analytical SQL) or Microsoft Fabric (unified SaaS platform). 55 cards on a topic that comes up often in professional Azure contexts.

55 cards
4
Data Governance / Purview

Microsoft Purview for Data Catalog, automatic classification and data lineage. Data Mesh and Data Products as an organizational paradigm. Governance challenges in a large Azure organization.

45 cards
5
Streaming / IoT / Event-Driven

Azure Event Hubs, Service Bus, Event Grid, Stream Analytics, Azure IoT Hub. Event-driven architecture, delivery guarantees, at-least-once vs exactly-once, and real-time processing patterns.

50 cards
6
Azure ML / AI / MLOps

Azure Machine Learning Studio, MLflow on Azure, AutoML, ML pipelines, production model monitoring, Responsible AI and Azure AI Services. ML industrialization on the Azure platform.

45 cards
7
Architecture Trade-offs

Developing an architect's mindset: how to weigh performance, cost, maintainability and scalability. Practical Azure architecture decision cases and a methodology for structuring a recommendation.

53 cards

Recommended Study Method

This guide is dense and technical. Allow 6 to 8 weeks of regular review for solid understanding. Unlike Azure Fundamentals, the goal is not to memorize service names — it is to understand the trade-offs.

1
Patterns and data lake (weeks 1-2)

Decks 1 and 2: Cloud Architecture Patterns then Data Lake & Lakehouse. These two decks establish the cross-cutting concepts the others build on. Don't move to deck 3 until you feel comfortable with Delta Lake and Medallion concepts.

2
Platforms and governance (weeks 3-5)

Decks 3 and 4: Databricks/Synapse/Fabric then Governance/Purview. Deck 3 is the most relevant in professional contexts — take time to understand each platform's use cases. FSRS will automatically schedule reviews of decks 1 and 2.

3
Streaming, ML and trade-offs (weeks 6-8)

Decks 5, 6 and 7: Streaming/IoT, MLOps, Trade-offs. The last deck is designed to be studied at the end of the guide — it puts all learned concepts into practice by forcing you to choose between valid architectures.

Frequently Asked Questions about the Azure Architecture Guide

What level of prerequisite is needed for this guide?

This guide targets profiles who already understand cloud fundamentals (Azure Fundamentals level or equivalent experience). It is not suitable for complete beginners on Azure. If you are new to Azure, start with the Azure Fundamentals guide (8 decks, 355 cards) before tackling this one.

Does this guide prepare for an official certification?

This guide is not aligned with a single Microsoft certification. It covers topics present in several advanced certifications: Azure Data Engineer (DP-203), Azure AI Engineer (AI-102) and Azure Solutions Architect (AZ-305). It is mainly designed to build Azure data architecture skills in a professional context — it does not replace official training or hands-on labs.

In what order should I study the 7 decks?

Recommended order: Patterns → Data Lake & Lakehouse → Databricks/Synapse/Fabric → Governance/Purview → Streaming/IoT → MLOps → Trade-offs. This order moves from the general to the specific and from the conceptual to the practical. The last deck (Trade-offs) is designed to be approached once the first 6 are mastered.

What is the difference between Databricks, Synapse and Fabric?

That is exactly what deck 3 covers (55 cards). In short: Databricks excels for complex data engineering and ML; Azure Synapse Analytics has historically been strong on analytical SQL and DWH; Microsoft Fabric is Microsoft's new unified SaaS platform. The choice depends on your existing stack, skills and priority use cases.

How is this guide different from memia's Data & AI cluster articles?

This guide covers data concepts (Data Lake, Data Mesh, governance, MLOps) from the angle of choices and implementation specific to Azure — which Azure service to use, and why. The Data & AI cluster (see "Further Reading") explains these same concepts in a vendor-neutral way, independent of any cloud provider. The two are complementary.

How long to complete this guide?

Allow 6 to 8 weeks with 20 to 30 minutes per day. This guide is denser than Azure Fundamentals — the 343 cards cover technical topics that often require progressive anchoring. The FSRS algorithm adapts the pace to your memory and schedules reviews automatically.

Start with architecture patterns

50 cards on core cloud patterns — the foundation of any Azure architect. Post-Azure Fundamentals level. Immediate access, no credit card required.

Get started for free

Further Reading