Is CI/CD for Fabric finally completely there?
This post addresses whether Microsoft Fabric's CI/CD capabilities have matured sufficiently for enterprise use. While Fabric's Deployment Pipelines exist, they lack flexibility for custom scenarios. However, recent developments—particularly Fabric's Git integration and the Fabric CI/CD toolkit—make the current state "close enough."
Key Implementation Strategy
At DW Data, we separated Fabric resources into three workspace concepts:
- Store (Lakehouses, Warehouses, EventHouses)
- Integration (Notebooks, Pipelines, DataFlows, Functions)
- Presentation (Reports, dashboards, apps)
This architecture enables cleaner boundaries, workspace-level RBAC, selective deployments, and feature branches pointing to shared pre-production data stores.
Strengths Identified
Several working capabilities we've leveraged:
- Automated workspace provisioning – no more manual clicks
- Environment promotion automation (feature → PPE → PROD)
- Lakehouse schema synchronization
- DevOps pipeline integration
- Strong support for PowerBI's new project format
Identified Gaps
Two significant limitations remain:
- Incremental deployments – Currently redeploys everything despite minor changes
- Service principal support for Azure DevOps – Only GitHub supports SPNs currently
Conclusion
Fabric's CI/CD is mature enough for enterprise implementation, though incremental deployments and better SPN support would complete the solution.