Your code as posted does not reproduce the problem, because you never call run_command.

My guess is that the command as constructed is wrong for some reason, and that makes the called program exit. I would print out the commands before they are run, and then check that the commands actually can run on their own, before blaming IO::Pipe.

sub run_command {
    my $user = 'gmon';
    my $system = shift;
    my $protocol = 'ssh';
    my $ssh_flags = "-l $user";
    my $command = "statvv -ni";
    my $space = " ";
    my $do_command = $protocol . $space . $ssh_flags . $space . $syste
+m . $space . $command;
    print "Running [$do_command]\n";
    my $cmd = IO::Pipe->new;
    $cmd->reader($do_command);
    return $cmd;
}
[download]

Also, why are you using IO::Pipe at all if you just want to run one command? You can use Perl to open a pipe-filehandle:

sub run_command {
    my $user = 'gmon';
    my $system = shift;
    my $protocol = 'ssh';
    my $ssh_flags = "-l $user";
    my $command = "statvv -ni";
    my $do_command = "$protocol $ssh_flags $system $command";
    print "Running [$do_command]\n";
    open my $fh, "$do_command |";
    return $fh;
}
[download]

... or you could just read the contents of the command directly:

sub run_command {
    my $user = 'gmon';
    my $system = shift;
    my $protocol = 'ssh';
    my $ssh_flags = "-l $user";
    my $command = "statvv -ni";
    my $do_command = "$protocol $ssh_flags $system $command";
    print "Running [$do_command]\n";
    my @results = `$do_command`;
    return @results
}
[download]