Fabric keeps the lake in sync as the source schema changes. The Schema Inspector compares the source schema with the current destination table schema, detects additions, deletions, and type changes, and lets you configure transform rules for column mapping.
Additive evolution
Fabric evolves the schema additively. It never retypes an existing column — the first batch wins for inferred types. New columns are added; a rename lands as a new column rather than being replayed.
Transform rules
You can configure per-column transform rules to map source columns onto the destination. A schema refresh can be triggered on demand, and the inspector shows a diff of what changed.
The schema guard
If a change is breaking, the schema guard refuses it rather than silently corrupting the lake. The runner holds — it doesn't crash-loop — so you can review the change and decide whether to apply a rule, evolve the destination, or stop the sync.
Why it matters