Databricks made multi-statement transactions generally available across Delta Lake and Apache Iceberg, expanding the range of traditional data warehouse workloads that can run directly on its Lakehouse platform.

The capability allows multiple data operations to succeed or fail together with atomic guarantees, reducing the need for separate warehouse systems and duplicated data. Databricks said the release is intended for complex ingestion, transformation, BI, application and operational analytics workloads requiring consistent updates across multiple tables.

Source: databricks.com