Delta table schema evolution: add, reorder, rename, widen types
Update table schemas with schema evolution
Databricks guide to explicit and implicit table schema changes, including DDL column operations and conflicts with concurrent writes and streams.
Based on
Update table schemas with schema evolution | Databricks on AWS
Tables support structural change as requirements shift. Supported change types listed are adding columns at arbitrary positions, reordering columns, renaming columns, and widening types under automatic schema evolution. Manual changes use ALTER TABLE without writing new data—for example ADD COLUMNS with optional comment and FIRST or AFTER position; nested fields use dotted paths, with element, key, and value path properties for structs inside arrays or maps. Nullability defaults to true for added columns.
The docs stress that schema updates conflict with all concurrent write operations and recommend coordinating changes. Updating a table schema terminates streams reading that table; processing continues only after restart per Structured Streaming production guidance. Implicit evolution via DML is mentioned alongside explicit DDL.
Schema is the shared contract of column names, order, and types. Controlled evolution keeps that contract aligned with producers and consumers; unmanaged concurrent writes or unrestarted streams break readers that still expect the prior shape. For governed semantic models, rename and widen operations are exactly the moments when physical schema and published meaning must stay synchronized.
Put this to work on CoreModels
Related connectors and recipes
Take the next step
Try CoreModels, talk with our team, or explore more resources.