Hi Team,

Currently Fivetran is automatically transforming columns data types for replicated Oracle database tables as per this documentation .

The mapping of NUMBER columns to FLOAT8 on Redshift destination is a bug with the fivetran data mapping when dealing with Monetary Values defined on oracle database as NUMBER data type which provides maximum precision and scale for the values stored.

Cause as per Redshift documentation, REAL, FLOAT and DOUBLE PRECISION are inexact types, meaning that some values are going to be stored as approximations. Hence, when dealing with Monetary values, the column data type should be stored on redshift with DECIMAL data type.

According to this, it would be great if Fivetran connectors provide the ability for users to select the data type of replicated columns when needed.

Thanks

Dina Elgendy