Does the choice of a single mapped task over 100 items versus 100 separate standard tasks have any impact from a pricing perspective?
09/23/2021, 9:26 PM
Each mapped task is a first class task with retries and observability so there is no billing distinction between mapped tasks and normal tasks. A mapped tasks of 3 items is the same as 3 independent tasks. If you have mapped tasks with thousand of elements, sometimes using a DataFrame directly can save a lot of money.
You only need tasks for stuff that can fail in general and maybe would benefit from a retry or the additional monitoring. Like, maybe you should make a task to apply a function on each row of a DataFrame rather than doing