it is much cheaper to order primers from your usual oligo supplier with the sequence NNNNNN than to buy hexamers from molecular biology suppliers who will sell you very small amounts. If your oligo supplier does not like NNNNNN then buy 4 oligos ANNNNN,
CNNNNN, GNNNNN and TNNNNN and it will still be very cheap. If this is for reverse transcription then I prefer random 9mers rather than 6mers
Paul is right and that works well for most applications, but in order to get the same coupling efficiency for each base at a given position (i.e. 25%A, 25%C,25%G,25%T), a special "hand mix" for an additional fee may be required depending upon how the oligos are synthesized. If not, your base representations will be skewed. There are also many companies that have random hexamers as stock items that are reasonably cheap.