My first thoughts were to use the system memory to load data file lines into a hash and load the DB records into an array (or visa versa) and check against each other. My hesitation on that was the memory load if the lists are large and more than one user runs this script

There's no need to load all of both datasets into memory.

If you load the names from the DB into a hash (which DBI will do for you in one hit--select names from table; fetchall_hashref), then you can process your file line by line.