Raid level 3
  SCSI or SATA (3 or more drives).

  Striped at byte level.

  Data distributed round robin at the byte level.
    High access speed.

    Each file distributed across all disk, only one file accessed at a time.

  Hamming code replaced by parity blocks on an extra drive.

  Parity protects against disk failure because failed disk can be IDed.
    Data errors of other types must be corrected with Reed-Solomon or some
    other in-line error recovery.

  Drives needed synchronization.

  Recovery simple.
  
  Efficient for very large files.

  Failed drive does not impede access.

  But recovery may take several hours.

  Single file access. 

  Parity drive must be tweaked for all writes so is overworked.