My first thought was:

sub mysort {
        my ($direction) = @_;

        ( $hash{$a}{$orderby} <=> $hash{$b}{$orderby}
          ||
          $hash{$a}{$orderby} cmp $hash{$b}{$orderby})
        *
        ($direction eq "DESC" ? -1 : 1);

}
[download]

This minimises duplication and cleans up the source code. But it still means evaluating the direction each time. All those wasted cycles! How about a closure?

my $orderby = "age"; # or name, dob, etc
my $direction = "ASC"; # or DESC
my $sortfunc = getsort($direction);

foreach my $id (sort $sortfunc keys %hash) {
  print "$id - $hash{$id}{$orderby}\n";
} # end-foreach

sub getsort {
        my ($direction) = @_;
        my $sense = $direction eq "DESC" ? -1 : 1;

        return sub {
                ( $hash{$a}{$orderby} <=> $hash{$b}{$orderby}
                  ||
                  $hash{$a}{$orderby} cmp $hash{$b}{$orderby})
                *
                $sense;
        } 
}
[download]

This keeps the code clean and keeps the operations to be done in each comparison close to the minimum, without using reverse.

Roger