It really depends on the machine that is running the code. Pandas will always have the entire thing loaded in memory, and while 600Mb is not a concern for our modern laptops running a single analysis at a time, it can get really messy if the person is not thinking about hardware limitations
I feel attacked