Build 2023: The Announcement
Microsoft Build happened in May 2023. Microsoft Fabric was announced. I predicted it in October 2022 and again in January 2023. I called the name in the October post.
I'm not writing this to say I told you so. I'm writing it because the announcement has real implications for your ADF investment, and I want to give you the clearest read I can on what it actually means — not what the press release says, but what it means for the pipelines you've been running for years.
What Microsoft Fabric Actually Is
Fabric is not a single product. It's a platform with six "experiences," each of which maps to something that existed before:
- Data Factory — ADF-equivalent pipelines plus Power Query dataflows. The orchestration experience.
- Data Engineering — Synapse Spark. Notebooks, Spark job definitions, managed Spark clusters.
- Data Warehouse — Synapse Dedicated SQL Pools equivalent. T-SQL analytics.
- Power BI — Power BI, unchanged in name but tightly integrated with the shared OneLake storage.
- Real-Time Analytics — Azure Data Explorer (Kusto). Event streams, time-series analytics.
- Data Science — ML experiments, models, and deployment. Azure ML's spiritual successor inside Fabric.
Six experiences. One workspace. One shared storage layer (OneLake, which is Delta Lake under the hood on ADLS Gen2). One governance model (Microsoft Purview integration baked in).
This is the consolidation I described in October 2022. The product family became a product.
What This Means for ADF Specifically
"Data Factory" in Fabric is ADF. Not ADF-equivalent — the pipeline authoring surface is identical. The same activity types. The same connector library. The same expression language. The same parameterization model. The same trigger types (schedule, tumbling window, event-based). If you know ADF, you know Fabric Data Factory. There is no new thing to learn for the core pipeline model.
This is intentional. Microsoft is not asking ADF customers to rebuild their knowledge. They're changing the container, not the engine.
What Happens to Standalone ADF
Microsoft has committed to continuing Azure Data Factory as a standalone Azure service. This is not end-of-life. If you're running ADF today, you can continue running ADF. The service will be maintained and supported.
The honest read: standalone ADF enters a maintenance posture. New feature investment goes to Fabric Data Factory. The connector library and core pipeline engine will stay current (shared infrastructure), but new Fabric-specific capabilities — deeper OneLake integration, native Fabric experience integration, workspace-level CI/CD — won't backport to standalone ADF.
This is fine. It's how Microsoft has always handled these transitions. SSIS still exists. SQL Server Integration Services is still a supported product. It's just not where new capability goes. ADF will follow the same path.
What the Timeline Looks Like
Fabric launched at Build 2023 as a preview. Microsoft has indicated GA in November 2023. That's the timeline for production readiness.
My recommendation: don't migrate production ADF workloads to Fabric during the preview period. Run parallel — keep production on ADF, experiment with Fabric Data Factory in non-production environments, understand the differences before committing. Validate the CI/CD model (which has changed), the IR setup process, and your specific pipeline types in Fabric before production migration.
You have time. Use it.
The Net Assessment
The ADF investment was correct. The skills transfer. The patterns transfer. The pipeline JSON transfers. The consolidation that was clearly coming has arrived, and it turns out that arriving at the consolidated platform from an ADF foundation is the easiest possible migration path.
If you've been building on ADF v2 patterns — parameterized pipelines, metadata-driven frameworks, CI/CD automation, Data Flows — you are well positioned for Fabric Data Factory. The transition is real, but it's not a rebuild.
There's more to cover: what's technically different in Fabric Data Factory, what the migration looks like in practice, what the metadata-driven pattern looks like in the new environment. I'll cover each in the coming months.
As always, if you need to make decisions before I get to those posts, reach out.