@tttpham the idea is to write the large file out in several parquet files that you can then upload individually to the new destination
There a Parquet reader will bring the data back into KNIME and handle them as a single file:
You can also make sure they have the same structure: