How would you make a large merge operation more memory-efficient when joining two datasets for AXA XL Insurance?
