break the files in smaller sorted ones and then use a merge sort to merge the files unifying duplicated entries when required.

I released Sort::Key::Merger recently on CPAN, it can mergesort data from several files efficiently:

use Sort::Key::Merger qw(filekeymerger);

my @last;
my $sorted = filekeymerger { (split)[0] } @file1_parts, @file2_parts;
while(defined(my $line=&$sorted)) {
  chomp($line);
  my ($id, $up, $down) = split ' ', $line;
  if (!@last or $last[0] eq $id) {
    push @last, $id, $up, $down;
  }
  else {
    print "@last $id $up $down\n";
    @last=();
  }
}
print "@last" if @last;
[download]