As an aside, I've removed your C-style loop, which is error-prone

Error-prone is in the eye of the beholder, and the C-style loop avoids the ugly $#array syntax.

Your data structures also scream for a hash, as using parallel arrays also is error prone

That would be a array of hashes then. But that has a severe drawback: it uses a lot of memory. Aggregates have a lot of overhead in Perl. The three parallel arrays only use three aggregates - regardless of the number of entities stored. Putting each entity in its own hash gives you an aggregate for each entity (that is, for each element of the original @name1). And that adds up. Fast.

Below is a program that shows the difference in memory usage . The hashes solution uses almost two and a half times the amount of memory.

#!/usr/bin/perl

use strict;
use warnings;
use Devel::Size qw 'total_size';

my $size = 100_000;
    
my (@name1, @name2, @percent);
my @big;

sub r_name {  # Make a random name.
    my $r = "";
    $r .= ('a' .. 'z')[rand 26] for 1 .. 3 + rand(5);
    return $r;
}

for (1 .. $size) {
    my $name1 = r_name;
    my $name2 = r_name;
    my $perc  = rand(100);

    push @name1, $name1;
    push @name2, $name2;
    push @percent, $perc;

    push @big, {name1 => $name1, name2 => $name2, percent => $perc};
}   

my $size_3 = total_size(\@name1) + total_size(\@name2) + total_size(\@
+percent);
my $size_1 = total_size(\@big);

printf "Three arrays:  %10d (%6.2f)\n",  $size_3, $size_3 / $size;
printf "One structure: %10d (%6.2f)\n",  $size_1, $size_1 / $size;
    
__END__
Three arrays:     9573279 ( 95.73)
One structure:   23724662 (237.25)
[download]