問題文
A model has to be fitted separately for each of 800 stores. Each store's data is small, and the 800 fits are independent of one another. Which construct lets Snowflake run these fits in parallel over the partitions of the data?
選択肢
- A Python user-defined table function that receives one partition at a time and returns rows, because the partitions can then be processed concurrently.
- A scalar user-defined function called once per row, because calling it 800 times is enough to cover the stores.
- A single stored procedure that loops over the 800 stores in sequence, fitting one store then the next, because a loop is the clearest way to express the work.
- An external function that receives all 800 stores at once, because the remote service can spread the work across its own pool of workers and return the 800 fitted models together.