See, I would have used something like ($data_hash{$index}) = $str =~ m/^(.+?)(?=\|)/. I wondered if it was faster to stop on the pipe with [^|] or use a lookahead:

use strict;
use Benchmark;

my $index;
my $str = "lol238923892382938|lol282812|asdfasdf|asdfasdfasdf";

timethese(5000000, {
    'stopper'   => sub { ($index) = $str =~ m/([^|]+)/ },
    'lookahead' => sub { ($index) = $str =~ m/^(.+?)(?=\|)/ },
    'splitter'  => sub { ($index) = split /\|/, $str },
});
[download]

The lookahead is technically faster on my machine, but not by enough to count as a victory. I'd be curious about other's results. Sadly, the splitter wins over the regulars by a similar (i.e. smallish) amount.