You make a good point, but I found myself wondering how the combination of substr, length, and eq would compare to a single anchored regex, so I did a benchmark. I'm convinced!

bannor:~/work/perl/monks$ cat 968838.pl
#!/usr/bin/env perl
use Modern::Perl;
use Benchmark qw(:all);

my $str = 'A.SERIAL_NUMBER||abcdefghijklmnopqrstuvwxyz';
my $ss = 'A.SERIAL_NUMBER||';
my $yay;

cmpthese( 10_000_000, {
    'substr' => sub {
        if(substr($str,0,length($ss)) eq $ss ){
            $yay = 1;
        }
    },
    'regex'  => sub {
        if( $str =~ /^\Q$ss\E/ ){
            $yay = 1;
        }
    },
});

bannor:~/work/perl/monks$ perl 968838.pl
            Rate  regex substr
regex  1312336/s     --   -65%
substr 3802281/s   190%     --
[download]