Yeah, that works too. And slicing the data in half is certainly a possibility. If you insert millions of rows, and you only have a single errors this may be more efficient than to do a loop one-by-one. Particularly, if you are lucky and the first half is good. (Then you know that you need to slice the second half already.)