Hi there,

In using the open source Fivetran terraform provider we noticed that schemas are considered a single entity from the provider, which means effectively that any change triggered (eg. adding a column, table) results in Terraform outputting the entire schema (each table+column) being deleted and re-created.

With small schema definitions this may not be an issue, however for longer schemas the output is huge and really difficult to read. This issue translates into the difficulty of understanding whether the changes made were correct, potentially leading to mistakes. 

Find attached an example of the output produced by the apply/plan when one single column is added to the schema for the "log" table in the "fivetran_log".