Does unRAID need a hardware controller to create a RAID array, or does it offer software RAID as an option?

 

UnRAID doesn't use ANY RAID controller.  It implements its fault tolerance entirely via software.

 

... and although it does implement fault tolerance, with either single or dual failure protection (i.e. either single or dual parity), it is technically not "RAID" => although clearly it uses a very RAID-like protection mechanism.  The key difference is it doesn't span the data across all disks.  While this results in a performance penalty, it has the advantage that you only need to spin up one disk to access any given file; and only need to spin up one disk plus the parity disk(s) to write new data.

 

 

 

Excellent. Thank you for the info. Is there a recommended quantity of disks to use? Something like at least 2, preferably 3+?

A common starting point with unRAID is two data drives and one parity drive.  You can then add data drives whenever more storage space is needed.  The newest version of unRAID, 6.2.x, supports up to two parity drives.  As you start increasing the number of data drives in your array it makes sense to add the second parity drive.

Thanks again

Excellent. Thank you for the info. Is there a recommended quantity of disks to use? Something like at least 2, preferably 3+?

 

The number of disks you choose to use is primarily a function of how much storage you need.

 

Assuming you want to take advantage of the fault tolerance, the minimum would be two disks -- one for parity, one for data.  You can then add additional data disks as needed to expand your storage capacity.

 

Note that the parity disk must be as large as any of the data disks => so be sure to consider what size disks you want to be able to use in the system, and buy that size disk for parity, even if you aren't doing to start with the same size data disk.

 

Also, as tdallen noted, you can have two parity disks if you want to be dual fault tolerant (i.e. you won't lose any data even if two disks fail).  Clearly this becomes more important as the number of disks gets larger -- but it's a very nice feature.  Of course, that also means you will be dedicating two disks to fault tolerance.

 

Finally, note that there is a maximum number of disks you can use, based on the license you buy.  The license types are noted here:  https://lime-technology.com/buy-it/

 

Note that the "attached devices" limit includes ALL disks in the system -- not just those you've assigned to the array.  So consider that when deciding which license to buy -- or you can just start with Basic and upgrade later for a small penalty ($9.00).  The maximum is 28 data drives plus 2 parity drives for a Pro license, which no limits on the total number of attached devices.

 

 

 

At least as important as the fact that some disks can be left spun down for file read/write accesses is the fact that each data disk contains its own individual complete file system so, unlike traditional striped RAID arrays, you don't lose everything even if you experience more failures than you have parity disks.

At least as important as the fact that some disks can be left spun down for file read/write accesses is the fact that each data disk contains its own individual complete file system so, unlike traditional striped RAID arrays, you don't lose everything even if you experience more failures than you have parity disks.

And another extremely useful feature that results from independent filesystems (no striping) is the ability to mix drive sizes.

... And another extremely useful feature that results from independent filesystems (no striping) is the ability to mix drive sizes.

 

Agree.  The only constraint (as I noted earlier) is that the data disks can't be larger than the parity disk(s)

 

... but the ability to mix sizes can be very useful both when initially building an array from whatever disks you happen to have handy; and later, even if you started out with all disks the same size, if you decide to "upsize" the array to larger disks, which allows you to do this incrementally.