I was just wondering how multiple "parity disks" would work, since it seems that doing so might defeat the purpose of having a parity.
Two parity blocks are needed for RAID-6; one is a standard XOR over the data blocks as in RAID-5, while the second involves math that's over my head.
The second type of parity allows for a recovery in a case where two data blocks, or a data block plus the standard parity block, are lost.
6And the main reason people want this is that disk sizes have gone up faster than disk speeds have gone up. So it takes longer to rebuild an array, even if you have a hot spare. That means the window in which two drives can fail has gone way up. Also, even if only one drive fails, because the amount of data you have to read off each drive to rebuild is so much more, the chance of a data error during rebuild is much higher. Having two parity drives solves all these issues. Oct 19, 2011 at 20:38
If you're referring to RAID 6 then it has two disks for parity for redundancy. It allows for 2 disks to fail and the array to survive, as opposed to one.
This gets more interesting once you have quite a big bunch of disks in an array. Gives you more chance to survive an additional drive failure than just having a hot spare.
As other have said, the multiple parity disks are full-blown duplicates of each other for extra redundancy.
But to really answer your question, the real reason behind these RAID models is large, cheap SATA drives. Unlike SCSI drives, you can plunk down $170 for a 2TB SATA drive (instead of $329 for a 450MB SCSI). The baggage that comes with SATA's high storage density is the heat and shorter lifetimes (or mean time between failures if you prefer).
Thats why we have RAID6...thats why we have multiple parity disks -- because the world has massive storage needs and is currently very short on money.
Today, using these disposable terabyte-sized SATA disk drives on mission critical application is only possible because of extra redundancy.