.. but what the problem is that when i run it i get the error message...uninitialized value in string concatenation ..

That usually occurs when you're putting an undefined variable into a string. Perl deals with this nicely and just puts a null string '' for you, but it's a warning that you probably have something wrong in your code. There's usually a line number along with that message -- check that part of your code.

I have a few suggestions for you that will probably help you in your quest to become a better developer.

• Be neat If you're going to write source code, you are writing for two audiences: computers (who don't care what it looks like) and humans (who appreciate regularly formatted code). Even before we start talking about use strict, we need to make sure this is clear.
• Be organized You need to lay out what the script is doing, and what your plan is for achieving that. A few comments will help that aim. No, you probably don't need to do that for a two line script. However, anything moderately complicated will benefit from this approach.
• Let Perl help you When you see a warning from Perl, that's a sign that you have missed something. If you clean up your code so that Perl can't find anything to complain about, the code may still not work, but at least it's doing the wrong thing with the right syntax.
• Perl is not like other languages Perl can help you do clever things. In your code, you have

    my @entries = $return->entries;
    foreach my $entry (@entries) {
    ..
  [download]

  So you called this function to get a list of entries, put it into an array, and in the very next step you start stepping through that array. That's a memory expense that you don't need. Instead,

    foreach my $entry ($return->entries) {
  [download]

  will do exactly the same thing.

Your approach to the problem makes sense -- you want to scan three servers and find all occurrences of the 1000 users. But you appear to be collecting that information as you go, rather than outputting it as you find it. Why is that?

I would suggest you use a hash of hashes (HoH) to group your users together -- so you'd have the names grouped together by first letter -- and then search as you've done on the same first letter, and check each entry to see if there's a match. The CPAN page on Net::LDAP says that you can use a callback function, so that might be a timesaver -- you can search for the users, specify a callback function and have that function stuff the matches into an array; once you return from the search, dump out the array of matches, and go on to the next letter. You can even limit the number of matches during your development, so as to not load up this Production machine.