The reasoning behind triple-parity RAID-Z
There was some discussion with colleagues and friends if we really need Triple Parity RAID. Adam Leventhal wrote an article about the need for Triple-Parity, the Math behind it and why there isn´t a generic n-Parity RAID at the moment. You will find the the article in his blog. As i wrote before, the question of the need for triple parity or n-parity condenses down to the probability of additional drives dying away (or delivering read errors) while syncing the new disc into the raid set. Just think about 4 TB hard disk (we already reached 2 TB) and large wide-stripe sets. With disks getting bigger and bigger, the time to resync the data gets longer and longer and thus you have a prolonging window of vulnerability for your data that gets more dangerous with the amount of disks in a raid-set.