thank you EVERYBODY for the responses. I didn't mention it before but I had tried optimizing things by using different types of fetches, binding, etc. but nothing seemed to help that much. However, I do have things working much faster now( even faster than it was on RHEL 4)!. I don't know why, but here is what I found thanks to tux's post. I do need the "limit 1" because otherwise, I don't get the most recent description, I get the oldest. However, for whatever reason, when I use desc AND limit 1 things are slloooww. Without "limit 1" things are fast... So, it is much faster for me to iterate through all of the rows of ids just to get the "latest" than it is for me to sort in descending order and get just 1. Hope that makes sense. Thanks again everyone. My code is much more optimized and I'm sure throughout the application there are things can be optimized and more secure so I am going to work on that. So, in short, still don't know why the same exact Perl code, with the same data, is much slower on RHEL 6 but my issue is resolved. So this is the code I ended up using and it is faster and seems to give me the correct results

my $dbh = $me->{DBHANDLE};
my %labels;

# Get all the different ids
my $sth = $dbh->prepare ("select distinct id from list_index");
$sth->execute;
$sth->bind_columns (\my $id);

my $sth2 = $dbh->prepare (qq;
    select description
    from   list_index
    where  id = ?
    order by rpt_key;
    );
$sth2->execute (0);
$sth2->bind_columns (\my $desc);

while ($sth->fetch) {
    $sth2->execute ($id);
    while ($sth2->fetch) {
        $labels{$id} = "$id - $desc";
        }
    }
$_->finish for $sth, $sth2;

return \%labels;
[download]