Data type matching between OLTP & DW

Hello All,

Is it generally considered a best practice to have the data types in your data warehouse tables (Facts & Dimensions) to match those of the OLTP system?

For instance in our OLTP system (MSSQL Server) we have many transaction dollar amount columns as a money data type (horrible, I know). I would prefer to bring this data into our DW as a decimal data type.

Similarly, there are many columns in our OLTP tables that are unicode (NVARCHAR) that have no need to be. That is to say there will never be unicode data stored in these columns. Rather than waste storage space needlessly in our DW, I would prefer to convert this data to non-unicode (VARCHAR).

Is this considered ok? Or is it best to keep data types the same between the systems?

Thanks,
Peter

Considering that most warehouses are sourced from multiple sources it is almost inevitable that the source and target data types (and lengths) will be different. It is expected. Change with confidence.

Besides, the MONEY data type is a decimal data type. Basically, it is a scaled integer, similar to most NUMERIC type implementations.

As B&L indicated, its par for the course.