The RAID GAME: Why RAID 6Over the past year or so, there has been increasing interest in the heretofore neglected disk array format known as RAID 6. Also known as dual-parity RAID, it addresses an important limitation on the more frequently implemented RAID 5; RAID 5 can't recover if two disks in an array fail at the same time. RAID 6 can sustain two simultaneous drive failures without data loss. A RAID 6 array is essentially a RAID 5 array with a second independent parity scheme. It stripes data on a block level across arrays like RAID 5, but uses a second set of parity codes for each stripe. If each checksum needs no more capacity than the data it protects on each member disk, a RAID 6 array with X member disks of user capacity requires X+2 member disks. RAID 6 writes data more slowly than RAID 5 because of the extra overhead. As a general rule, RAID 6 is more costly to implement because it requires special, very smart controllers as well as more drives, a minimum of four. This has given the IT community with its limited budgets pause over the years. But the days of SATA drives are here and serial-attached SCSI (SAS) drives are virtually here.
SATA drives, rightly or wrongly, are perceived as lower performance drives and in some cases theyare considered less reliable. Without taking sides in that dispute, lower cost SATA drives might help ease budgetary strains. There is also some performance degradation on random writes. But in data centers where performance is less important than capacity and spindle failures, RAID 6 might well be worth the investment. Some might ask: If the system must make two backups, how will the functions be impacted? Aside from a random write efficiency of about 70% of RAID 5, the other RAID 6 functions are fully equivalent to RAID 5.
| 
Chinese Traditional