Instead of creating the suffixes, you can directly sort the indices based on a call to substr in sort. My machine does this in less than 2 minutes for a random string of length 200,000.

use strict;
use warnings;

sub suffix {
    my $string = shift;
    my @array = sort { substr( $string, $a ) cmp substr( $string, $b )
+ } 0..length($string)-1;
    $_++ for @array;
    return @array;
}

my $n = 200000;
my $input = join '', map { ('a'..'z')[rand(26)] } 0..$n-1;
my @result = suffix $input;
print "@result\n";
[download]