There seems to be some common misconceptions regarding RAID (redundant array of independent disks) and what Unraid offers in its array with optional parity protection.

I'll go over some of the basic RAID types and explain the differences in Unraid afterwards.

RAID levels

Photos from Wikipedia


Minimum drive count : 2
Highly performant striping of data across all the disks in a set. However, no data protection with this type so any disk failure destroys the array.


Minimum drive count : 2
Mirrored data with no striping. Every bit is written across both drives at the speed of the slower drive. In case of drive failure, the array degrades but the remaining disk can still be used.


Minimum drive count : 3
Striping plus parity. Data is split evenly between at least two drives with parity being written to the third. The parity is distributed among the disks.
This provides a good balance of performance as well as redundancy but the trend with larger drive sizes lately is for another disk to fail during resilver due to the very large disk contents being restored.
If a second disk fails during resilver, the array is lost.


Minimum drive count : 4
Striping plus double parity. 2nd level of disk backup. Extremely similar to raid 5 just with an added redundancy. Useful in the modern era with the larger drives as a multi disk failure is less likely without a catastrophic event. If a third disk fails during resilver, the array is lost.

Raid 10

Minimum drive count : 4
Stripe of two or more RAID 1 arrays. Presents as one logical volume. Either drive in each RAID 1 array can fail without affecting the overall volume. If both RAID 1 arrays fail during resilver, the array is lost.

There are a few other uncommon or legacy RAID types but the above are by far the most common.


Unraid is not RAID.
Minimum drives : 1
Minimum drives with parity : 2

Unraid array without parity works extremely similar to JBOD (Just a Bunch Of Disks). It presents as a single volume to the operating system except that data is not striped across disks. Each file is written in its entirety to a single disk on the array. The benefit to this is that if a single disk fails all the other disks and data on them are completely recoverable without any special software. Only the data on missing disk is lost. This can be especially desirable with certain types of replicable media.

A dedicated parity disk allows the array to do some complicated math to emulate or recover any missing or failed drive. The drawback is that parity is calculated based on the bits of every disk. This means writing to the array is slower than a direct copy since it's simultaneously reading the matching bit address on all other disks in the array.

Parity drives are dedicated drives. They hold no data of their own but can help your system to recover and rebuild missing data from an array. The number of parity drives is the number of drive failures your system can tolerate before data will be lost. They key here is that you are only losing the data on that drive not the entire array.


Both styles have their benefits and drawbacks based on storage efficiency and performance. None of them conform to a true 3-2-1 style backup with the RAID 1 being the closest. Additionally I have not gone into the ZFS raid enhancements to the standard RAID levels. Suffice it to say that a RAID 5 roughly equals RAID Z1, and RAID 6 roughly equals RAID Z2.