Pipelines for Snowflakes
Pipelines in Snowflake are built from streams, tasks and dynamic tables. A stream records row changes on a table, a task runs SQL on a schedule or when triggered, and a dynamic table keeps a query result up to date from a declared target lag. Together they support incremental transformation without an external scheduler.
Stage: Transformation · Part of the Snowflake roadmap · Last reviewed 19 November 2026
What to learn in Pipelines
- Create a stream on a table and read inserted, updated and deleted rows
- Schedule a task that merges stream data into a target table
- Chain tasks into a dependency graph with a root task
- Define a dynamic table and set its target lag
- Choose between tasks with streams and dynamic tables for a transformation
Pipelines interview questions
What does a Snowflake stream store?
A stream stores an offset into a table's change history, not a copy of the data. Querying it returns the rows changed since that offset, with metadata columns showing the action. The offset advances when a DML statement consumes the stream in a transaction.
How does a dynamic table differ from a materialised view?
A dynamic table is defined by a query and refreshed to meet a target lag, and it can use joins and complex logic that Snowflake manages incrementally where it can. Materialised views have tighter query restrictions and refresh automatically on a single base table.
Where Pipelines fits in the Snowflake roadmap
Pipelines belongs to the Transformation stage (intermediate level). Build pipelines.
Other units in this stage
Back to the full Snowflake roadmap