This script takes the name of the file and the number of sequences you want and outputs that number of randomly selected sequences.

It is simple, fast and should handle any size of input file with minimal memory usage:

#! perl -slw
use strict;
use List::Util qw[ shuffle ];

$/ = '>';

open FASTA, '<:raw', $ARGV[0] or die $!;
my @seqPosns;
push @seqPosns, tell( FASTA ) while <FASTA>;

@seqPosns = shuffle @seqPosns;

for ( @seqPosns[ 0 .. $ARGV[ 1 ] // 10 ] ) {
    seek FASTA, $_, 0;
    my $seq = <FASTA>;
    chomp( $seq ); chop( $seq );
    print '>', $seq;
}
close FASTA;

__END__

C:\test>988096.pl C:/dell/test/LCS/bioMan.fasta 2
>af418682
TTCCACAACTTTCCACCAAGCTCTACAAGATCCCAGAGTCAGGGGCCTGTATTTTCC
TGGGTCTTTTGGGCTTTGCCGCTCCATTTACACAATGTGGTTATCCTGCATTAATGC
ACTTCTTTCCTTCAGTACGAGATCTCCTAGATACCGCCTCAGCTCTATATCGGGAAG
TCAAACAATCCAGATTGGGACTTCAACCCCATCAAGGACCACTGGCCACAAGCCAAC
>ab033557
CTCCACGACATTCCACCAAGCTCTGCTAGATCCCAGAGTGAGGGGCCTTTACTTTCC
TGGGTCTTTTGGGCTTTGCTGCCCCTTTTACACAATGTGGCTATCCTGCCTTAATGC
ACTTCTTTCCTTCCATTCGAGATCTTCTCGACACCGCCTCTGCTCTGTATCGGGAGG
TCAAACAATCCAGATTGGGACTTCAACCCCAACAAGGATCAATGGCCAGAAGCAAAT
>x97850
CTCCACAACTTTCCTCCAAACTCTTCAAGATTCCAGAGTCAGGGCCCTGTACCTTCC
TGGGTCTTTTGGGGTTTGCCGCCCCTTTCACGCAATGTGGATATCCTGCTTTAATGC
ACTTTTTTCCTTCTATTCGAGATCTCCTCGACACCGCCTCTGCTCTGTATCGGGAGG
TCAGAAAATCCAGATTGGGACCTCAACCCGCACAAGGACAACTGGCCGGACGCCAAC
[download]