It may be worth considering dumping the milcar file into an SQLite database, using Perl to get your data from file2, and then using the modules DBI & DBD::SQLite along with SQL's native functions to find db-matches against file2's fname, lname.

That's not to claim I'm sufficiently expert with dbs to guarantee that this is the best approach but -- based on limited experience with (mostly) toy-applications -- it looks pretty plausible to me and DBI and DBD::SQLite will make the learning curve much shallower while the SQL capabilities can handle most of your search, sort and write requirements.