Migration Playbooks

Opinionated, step-by-step conversions from Azure Synapse, Databricks, Snowflake, and SSIS into Microsoft Fabric.

Each playbook follows the same arc: inventory → map → move a slice → validate → cut over → decommission. They assume you have read OneLake & Delta Lake and CI/CD & Git.

Still deciding whether to move at all, not just how? The comparison card below is a good place to start.

From Azure Synapse Analytics

Dedicated/serverless SQL pools, Spark pools, pipelines, and Synapse Link into Fabric.

From Databricks

Notebooks, jobs, Unity Catalog, and DLT into Fabric lakehouses + pipelines.

From Snowflake

Warehouses, roles, tasks, and SnowSQL into Fabric Warehouse + OneLake.

From SSIS

Control flow, data flow, and SSISDB into Fabric Data Pipelines + Dataflows.

Fabric vs Databricks

An honest comparison — billing model, catalog, BI integration — before you commit to migrating.

Fabric vs Snowflake

Storage openness, compute sizing, and governance — before you commit to migrating.

Universal pre-work

Stand up the landing zone

A Fabric capacity, a dev workspace connected to Git, one lakehouse, one warehouse. Nothing production yet.

Inventory the source

Enumerate every job, table, and dependency. Tag each: keep / retire / rebuild. Roughly a third of most estates is dead code — do not migrate it.

Pick a thin vertical slice

One source table → one transform → one consumable output. End to end. This proves the pattern and your CI before you scale out.

Define "validated"

Row counts match, a set of aggregate checksums match, and a business user signs off on one report. Write these down before you start moving data.

Run old and new in parallel for at least one full business cycle. The migration is not done when the new pipeline runs — it is done when you have deleted the old one.