Hi Eugene,

One more thing…

I double checked error handling with my colleague who knows far more about data load than I do.

Error handling in binary case may be more difficult. Namely, when loader reports that there was an error in a given row (with row number), it’s harder to find that row in binary case.

On the other hand, my colleague confirms that binary case should be faster.

Thus, I guess that it’s first of all worth checking the speed in binary case. Then, if it’s significantly better, we can go back to the topic of error handling.

Please let us know whether it makes sense for you.

Thanks and greetings!

Dominik

Hi !

It is more troublesome to prepare a binary load file, usually a dedicated program/script is necessary to create binary load files. Fortunately, many formatting errors can be check during preparation of binary files. Then, problems like unexpected field delimiters should not appear during the load - in the case of the binary format the loader just takes N bytes as the string content and does not care what is inside.