If you run a select on every table ordered by the primary key you can compare the rows without loading the full tables in memory.

For instance, supposing the primary key my_key is of some string type:

my $sql = <<EOSQL;
  select my_key, foo1, foo2, ...
    from bar
    order by my_key
EOSQL

my $sth1 = $dbh1->prepare($sql);
my $sth2 = $dbh2->prepare($sql);

unless ($sth1->execute and $sth2->execute) {
  die ...;
}
my $row1 = $sth1->fetchrow_arrayref;
my $row2 = $sth1->fetchrow_arrayref;

while ($row1 and $row2) {
  if ($row1->[0] lt $row2->[0]) {
    print "$row1->[0] only in db 1\n"
    $row1 = $sth1->fetchrow_arrayref;
  }
  elsif ($row1->[0] gt $row2->[0]) {
    print "$row2->[0] only in db 2\n"
    $row2 = $sth2->fetchrow_arrayref;
  }
  else {
    # a better equality check for rows could be needed
    # if the db software is different or if the db
    # schemas are not identical:
    if ("@$row1" eq "@$row2") {
      print "row $row1->[0] differ\n"
    }
    $row1 = $sth1->fetchrow_arrayref;
    $row2 = $sth2->fetchrow_arrayref;
  }
}
while ($row1) {
  print "$row1->[0] only in db 1\n"
  $row1 = $sth1->fetchrow_arrayref;
}
while ($row2) {
  print "$row2->[0] only in db 2\n"
  $row2 = $sth2->fetchrow_arrayref;
}
[download]