Useful Architectures

If you see this on github.com, please visit https://jugi92.github.io/useful_architectures/ for better visibility of the graphs.

This is a collection of useful architecture drawings, which solve one or multiple potential usecases, including comments or alternatives.

AI

Azure AI Landscape

Full view with Layers Azure AI Landscape

Analytics

Fabric

Full view with Layers Fabric

Fabric Assets animated

Fabric Assets

Fabric Hierarchy

Fabric Governance Types

Fabric Security & Networking

Fabric Data Access Management

Aligning goals with different speeds

Fabric with Customer Managed Key / Bring your own Key

Fabric Asset Table

• Workspace
  • Lakehouse
    • Folders
      • Files
    • Tables
  • Warehouse

Migrate Azure Modern Data Platform to Microsoft Fabric Intelligent Data Platform

• Synapse -> Fabric Overall Migration Guidance
  • Workspace -> One or more Fabric Workspaces (Multiple Dev / Feature Workspace, 1 Integration, 1 Production)
  • Data Explorer Database -> Fabric Real Time Database
  • Lakedatabase -> Lakehouse
  • Linked Services
    • Azure Blob Storage -> Migrate to OneLake OR Shortcut in Lakehouse
    • Azure Data Lake Storage -> Migrate to OneLake OR Shortcut in Lakehouse
    • Azure Data Explorer -> Migrate to Fabric Real Time Database OR Fabric Real Time Database Shortcut (Follower Database) OR Use connector in pipeline or Data Flow Gen2
    • SAP CDC Connector -> Mount Azure Data Factory into Fabric and keep using Mapping Data Flow
    • Key Vault
      • in pipelines & mapping data flows-> credentials are stored in connection via Cloud Data Sources
      • in Spark -> use mssparkutils
    • other
  • Synapse Link for Cosmos DB -> Mirroring via Fabric Mirroring for CosmosDB
  • Pipeline -> Fabric Data Pipeline
    • File Triggered pipelines -> Storage Event Trigger via Data Activator
  • Mapping DataFlow -> no direct successor, rewrite to Fabric Dataflow Gen2 OR rewrite to Fabric Notebook
  • Spark -> Fabric Data Engineering Experience (Migrating from Azure Synapse Spark to Fabric)
    • Notebook -> Fabric Notebook (consider Repo)
    • Spark Pool -> Fabric Spark Pool (consider Repo)
    • Spark Configuration
    • Spark Libraries
    • Job definition
  • dedicated SQL Pool -> Fabric Warehouse OR Fabric Lakehouse (see Migration: Azure Synapse Analytics dedicated SQL pools to Fabric, DWU Estimation)
  • Serverless SQL Pool -> Fabric Lakehouse SQL Analytics Endpoint
  • Integration Runtime
    • Azure Runtime / Azure Auto Resolve Runtime -> plain Fabric Compute Capacity
    • Self hosted Integration Runtime -> VNet Data Gateway OR On-premises Data Gateway
  • Git configuration -> Migrate to Fabric Workspace Git configuration (previously 1:1 relation between workspace and repository, now 1:1 relation between workspace and branch)
  • Azure DevOps Deployment Pipelines -> Azure DevOps Deployment Pipelines + Fabric Deployment Pipelines
• Power BI -> Fabric
  • P SKU to F SKU
  • Power BI Workspaces -> Keep as PBI Workspaces
  • Migrate Dataflow Gen1 to Gen2
• Blob Storage -> Migrate to OneLake (az copy) OR keep and Shortcut in Lakehouse
• ADLS -> Migrate to OneLake OR Shortcut in Lakehouse
• Azure Data Explorer -> Fabric Real Time Database OR Fabric Real Time Database Shortcut (Follower Database)
• Azure Cosmos DB -> keep as is, run analytical workloads via Fabric Mirroring for CosmosDB
• Azure Functions -> keep as Azure Function, adapt to Fabric
• Azure Event Hub -> keep as Azure Event Hub, integrate into Eventhouse via Eventstream

Synapse Analytics

Full view with Layers: Analytics-with-Azure Synapse

If you have an additional one or more insights on the existing ones, feel free to submit a Pull Request.