Implement Sampling > Random as Engine:In-Database(SQL) for Snowflake

ecerulm

Currently if I select Sampling method: Random (approx. ratio) or Random (approx. nb. records) the only allowed engine is DSS which will require downloading the input dataset to dss.

It's possible to do sampling at the Snowflake side, with https://docs.snowflake.com/en/sql-reference/constructs/sample

For Random(approx. nb.records) I believe it would be as easy as generating the following SQL

select * from input_table sample row (10 rows) seed (99);

ecerulm

SAMPLE/TABLESAMPLE is supported in many databases:

* Postgres

* teradata

* MySQL

* Google BigQuery

* Microsoft sql server (Transact SQL)

* Oracle