FabricOpsFlow

A CI/CD framework for end-to-end automation of Microsoft Fabric projects, integrating Azure DevOps Pipelines, Service Principal authentication, and automated deployment powered by pyFabricOps.

What It Is

FabricOpsFlow automates the entire lifecycle of a Microsoft Fabric project — from infrastructure provisioning (workspaces, lakehouses, pools, environments) to continuous deployment of artifacts (notebooks, semantic models, reports, and more) via Azure DevOps Pipelines.

Key Features

• Full environment initialization: An interactive Jupyter Notebook that provisions workspaces (DEV/PRD), lakehouses in a medallion architecture (bronze, silver, gold), Spark pools, environments with pre-installed libraries, Variable Libraries, and Azure DevOps Git integration — all via REST API.
• Automated deployment with 3 modes:
  • selective — deploys only changed items (based on git diff)
  • specific — deploys manually specified items
  • full — deploys all items in the repository
• Automatic parameter switching between environments: Replaces workspace IDs, lakehouse IDs, and connection strings in notebooks and semantic models when promoting from DEV to PRD.
• Automatic report conversion: Converts report definitions before deployment to ensure compatibility with the target workspace.
• Secure authentication: Credentials stored in Azure Key Vault and injected via Variable Groups in Azure DevOps.
• Permission management: Automatic role assignment (Admin/Owner) for users and service principals on workspaces and connections.

Architecture

| Component | Description | |---|---| | support/init_fabricopsflow.ipynb | Initialization notebook that provisions all infrastructure | | azure_pipelines/deploy.yml | Azure DevOps CI/CD pipeline | | scripts/deploy.py | Deployment script that orchestrates publishing items to workspaces | | scripts/utils.py | Git utilities for detecting changed items | | src/ | Directory with versioned Fabric artifacts (notebooks, semantic models, reports, etc.) |

Workflow

1. Developers work in feature workspaces FEAT, connected to feature/* branches
2. Pull requests, once approved and merged into the develop branch, trigger deployment to the DEV workspaces
3. Once features reach a new deliverable, pull requests to the main branch trigger deployment to the PRD workspaces
4. The pipeline detects changed items, performs parameter switching between environments, and deploys to the corresponding workspace
5. Reports undergo definition conversion before deployment

Prerequisites

• Microsoft Entra: App Registration with permissions for Power BI, Azure DevOps, and Key Vault
• Azure Key Vault with Service Principal credentials
• Azure DevOps: Organization, project, and repository configured
• Microsoft Fabric: API permissions enabled for the Service Principal security group

See the initialization notebook for detailed instructions.

Tech Stack

• Microsoft Fabric (Workspaces, Lakehouses, Notebooks, Semantic Models, Reports, Variable Libraries, Spark Pools & Environments)
• Azure DevOps Pipelines (CI/CD)
• Azure Key Vault (Secrets Management)
• Microsoft Entra / Service Principal (Authentication)
• Python + pyFabricOps (SDK)

License

MIT © 2026 Alison Pezzott