Depending on how flexible you need this to be, you might consider using an array of hash references, such that each row would be represented as a reference to an associative array with the column names as keys. Then you could sort the array of rows according to a particular key with a comparison subroutine:

foreach my $this_record (sort byname @record_set) {
  # do something
}

sub byname {
  $a->{'NAME'} cmp $b->{'NAME'}
}
[download]

See pages 789-793 of Programming Perl (3rd Edition)