Pool and Duplication method

[AlexW]

I have used the Drivepool trial version, and I have been unable to determine what method it uses for pool management and file duplication.

Does Drivepool use any of the common RAID solutions (0, 0+1, RAID 5 etc) or does it use an entirely proprietary solution?

[VapechiK]

yes.  1x means no dupes.  1 complete file only.  2x = 2 dupes.  2 identical complete files.  3x = 3 etc.  all (each) on separate physical drives (assuming enough drives are present).  DrivePool will NEVER split files.  it may/will split folders across different underlying drives to achieve whatever balancing ratio you have set up using the various available plug-ins.  the 'drive usage limiter' plugin (called 'file placement limiter' in the manual, maybe it was renamed) will allow you to choose which discs hold which types (duplicated/unduplicated or both).  'file placement rules' ensure certain type files go to certain discs/directories.

parity is a long requested feature for DrivePool.  as a long time reader of these forums i can say with some confidence it will NOT be included as an option.  many many users report success using SnapRAID for parity.  i don't use it so i have no guidance there.  my pool has 6 spinners and no duplication since i have backups and backups of some backups lol

so i would suggest RTFMs and the forums, even running a non-production test pool should be useful.  here are some links to get you started...

 

 

https://stablebit.com/Support/DrivePool/2.X/Manual

https://stablebit.com/Support/DrivePool/2.X/Manual?Section=File Protection
 

https://wiki.covecube.com/StableBit_DrivePool_Knowledge_base

 

[Shane]

None of the above. It creates hidden root folders, or "poolparts", on basic (not dynamic) NTFS-formatted volumes chosen by the user and collectively presents those folders as a single NTFS-style virtual drive and basic volume, or "pool". You can have multiple pools, you can even have pools of pools, and there is also ReFS support. If you're familiar with Linux, it's a "union" virtualization similar to mergerfs but (amongst other things) runs at the system level rather than in userspace.

Duplication is handled via multiple instancing of a file on the poolparts; there is no "original" vs "backup", there's just the same file existing on multiple volumes.

For example if you created a new pool, let's call it "P",  using six drives each formatted as a basic NTFS volume (they don't have to be the same size), let's call them "A" through "F", set 3x duplication for all files, and saved a file "example.txt" to the root folder of the new pool, it would look like this at the file system level (where guidstring is a alphanumberic identifier that uniquely identifies each poolpart):

P:\example.txt

A:\PoolPart.guidstring\example.txt

B:\PoolPart.guidstring\example.txt

C:\PoolPart.guidstring\example.txt

D:\PoolPart.guidstring\

E:\PoolPart.guidstring\

F:\PoolPart.guidstring\

Basically you just work in the pool drive and DrivePool does its thing in the hidden poolpart folders background; you normally never need to manually deal with the latter unless something has gone wrong (e.g. your old PC's mainboard died, you can't connect to the internet to download DrivePool to your new PC and for whatever reason you need to get a file from the pool right away) or you're using DrivePool in conjunction with other storage management software (such as SnapRAID for additional file integrity/recovery features).

There are a few advanced features of NTFS that it doesn't perfectly emulate that some programs complain about (e.g. Microsoft's OneDrive doesn't like to be installed on a pool), but otherwise IMO it's pretty great.

[AlexW]

Does this mean that files are duplicated equally on each drive? IE, a  4 GB video file stored in a pool will take up 4GB on 3 of the drives described above (and therefore taking up 12GB of practical drive space), or will it be split up with parity across the drives?

[lvau]

There is no parity in Drivepool, only mirroring of files. So your first example is what will happen.