Unless I'm overlooking something, you're overcomplicating it

use strict;
use warnings;

my %clean_data;
chomp (my @data_array = <DATA>);


for my $data ( @data_array)
{
    my ($server, @info) = split (/,/,$data);
    $data = \@info;
    push @{$clean_data{$server}}, $data;
}

#use Data::Dumper;die Dumper\%clean_data;
print
    $_,
    $/,
    map { "\t@$_$/"} @{$clean_data{$_}}
    for
        sort
            keys %clean_data;

__DATA__
server_A, Perl, UK
server_A, Word, UK
server_A, Outlook, UK
server_A, Excel, UK
server_B, Reuters, NL
server_B, TradeXL, NL
server_B, Thompsons, NL
server_B, Bloomberg, NL
server_B, Tibco, NL
server_c, BasketLink, USA
server_c, Evolution, USA
server_c, Lotus, USA
server_c, TIB, USA
server_A, Python, UK
[download]