The Iceberg STRING column type is mapped to Teradata VARCHAR character set UNICODE. Iceberg STRING column values can be large and does not have a maximum column length defined in the table schema. This makes it harder for TDOTF engine to do the datatype mapping with correct max column length.
To minimize data loss, by default, STRING columns are mapped to maximum supported VARCHAR UNICODE size of 32000. Using this default size of 32000 for smaller strings has shown to have a performance impact especially when used in a SORT BY or GROUP BY clause. Future releases of TD OTF will allow the default VARCHAR size of a Iceberg STRING to be tunable through a NVP setting - this can be used when the lengths of the STRING columns in the OTF dataset is known and can lead to better performance.