Default warehouse for Snowflake Notebooks (Pending)¶

With this behavior change, Snowflake Notebooks users can now specify separate warehouses for running the notebook kernel (Python code) and executing SQL queries in the notebook. The notebook warehouse remains active for the duration of the notebook session, while the SQL warehouse is utilized only on demand.

This separation enables users to assign a smaller warehouse for the kernel while optionally assigning a larger warehouse to handle heavy SQL queries. When creating a notebook, users can choose to specify two separate warehouses or use a single warehouse for running both the Notebook kernel and any pushdown SQL.

By default, the notebook warehouse is set to SYSTEM$STREAMLIT_NOTEBOOK_WH. However, users can specify a different warehouse at the time of notebook creation by choosing one from the dropdown list. After notebook creation, users can choose a different warehouse from the notebook settings.

Before the change:

CREATE [ OR REPLACE ] NOTEBOOK [ IF NOT EXISTS ] <name>
  [ FROM '<source_location>' ]
  [ MAIN_FILE = '<main_file_name>' ]
  [ COMMENT = '<string_literal>' ]
  [ QUERY_WAREHOUSE = <warehouse_to_run_nb_and_sql_queries_in> ]
  [ IDLE_AUTO_SHUTDOWN_TIME_SECONDS = <number_of_seconds> ]

Copy

After the change:

CREATE [ OR REPLACE ] NOTEBOOK [ IF NOT EXISTS ] <name>
  WAREHOUSE = <notebook_kernel_warehouse_name>
  [ FROM '<source_location>' ]
  [ MAIN_FILE = '<main_file_name>' ]
  [ COMMENT = '<string_literal>' ]
  [ QUERY_WAREHOUSE = <warehouse_to_run_sql_queries> ]
  [ IDLE_AUTO_SHUTDOWN_TIME_SECONDS = <number_of_seconds> ]

Copy

A new parameter, WAREHOUSE, has been introduced as a required parameter to specify the warehouse used to run the Notebook kernel and Python code. If this parameter is not explicitly set, it defaults to the value of the schema-level parameter DEFAULT_STREAMLIT_NOTEBOOK_WAREHOUSE, which determines the default warehouse to be used.

For details, see Default warehouse for Notebooks.

Ref: 1887