I am setting up an ETL process and small data warehouse for querying the data in a few different dimensions. One issue is that the schema for the objects can mutate over time - mainly that some fields will be added and some removed. What are some methodologies or approaches for handling this?
(Note the source data is managed as EAV, which is hard to query and it's slow - hence the DW approach in the first place).
What seems like a naive approach (which can often be the best) is to simply add columns over time to the destination shapes in the warehouse, and backfill existing objects with some query-valid placeholder data.
This is a little outside my domain expertise though, so looking for input.