Question with execution times using custom Integration Runtime in Azure Data Factory

Question

I’m trying to find the best configuration for my Integration Runtime to process a file with over 2 million registers.

I have two different IR. The first one has 4(+4)