The same value will always result in the same HashRow, that's why is said "xxx must be a column which is (close to) unique".
If it's the logical Primary Key of this table you're not going to load the same row a second time.
But just go for the Derived Table with RANDOM, it's easier to write and there's a big advantage:
The size of the RANDOM is always 4 bytes while the PK might be much more bytes, thus requiring more spool space.
Dieter
The same value will always result in the same HashRow, that's why is said "xxx must be a column which is (close to) unique".
If it's the logical Primary Key of this table you're not going to load the same row a second time.
But just go for the Derived Table with RANDOM, it's easier to write and there's a big advantage:
The size of the RANDOM is always 4 bytes while the PK might be much more bytes, thus requiring more spool space.
Dieter