James Serra ponders when it makes sense to add structured data to a data lake:
I would not say it’s common place to load structured data into the data lake, but I do see it frequently.
In most cases it is not necessary to first copy relational source data into the data lake and then into the data warehouse, especially when keeping in mind the effort to migrate existing ETL jobs that are already copying source data into the data warehouse, but there are some good uses cases to do just that:
There are some good reasons in here, so check them out.