Backing Up DrivePool using CloudDrive and Backblaze B2

[digimon]

I recently upgraded my home server and moved away from Windows 10 to Windows Server 2022. To my dismay, my personal Backblaze account does not work with Windows Server. After further research, it seems the solution is to upgrade to Backblaze B2. But I quickly realized that Backblaze B2 with Cloudberry also does not work to backup DrivePool drives because it is considered a "Volume Shadow Copy". I think I have a solution that might work, using drivepool with clouddrive and Backblaze, but I wanted to confirm with the Stablebit user community. My current drivepool configuration consists of four 3TB hard drives that are duplicated four times. My thought was to create a fifth 3TB CloudDrive that I would pool with my existing drives and change the duplication to 5 times. This way I would have 4 physical 3tb harddrives and one virtual 3tb harddrive. The virtual drive should automatically sync with backblaze B2. In case of a failure, I could restore the clouddrive in a new machine.

My second question is whether the 3TB clouddrive actually takes up 3tb of space in backblaze. If I only have 1.5tb of files, will I be charged for the 1.5tb of usage or the full 3tb drive.

Let me know if this though process is flawed or if it should work. Thanks for your help!

[Shane]

Are you familiar with the VHD, VHDX, etc type of file formats that allow a virtual hard drive to be stored as a single file whose actual size expands with the used space of the virtual hard drive it represents?

Clouddrive does something similar but instead of a single file it uses multiple files (called chunks, with each chunk roughly equivalent to a number of in-use sectors of the virtual drive).

If the clouddrive is kept on local storage (i.e. a physical drive pooled alongside the existing physical drives physical drive or replacing one of them) and you're using backblaze's software to back that up, then it's only going to back up the 1.5TB of the files in the clouddrive.

If the clouddrive is kept on blackblaze, then there will also be overhead from the virtual drive metadata plus the cost of any "sectors" that have been used but not (yet) cleaned. Also if clouddrive data duplication (not to be confused with drivepool duplication) is enabled when the clouddrive is created then the storage requirements will be doubled.

I hope that's clear?

P.S. If your pool consists of four physical drives with 4x drive duplication enabled (i.e. every drive is an exact copy of every other drive) have you looked at whether it's suitable for your situation to have the B2 backup software use the poolpart folder on any one of the drives (which are VSS-compatible) as its source instead of the virtual pool drive?

[digimon]

18 hours ago, Shane said:

P.S. If your pool consists of four physical drives with 4x drive duplication enabled (i.e. every drive is an exact copy of every other drive) have you looked at whether it's suitable for your situation to have the B2 backup software use the poolpart folder on any one of the drives (which are VSS-compatible) as its source instead of the virtual pool drive?

Hmm... I had not thought of backing up the actual physical drive since I have them hidden with no drive letters. And yes... I have 4x drive duplication enabled and the drives should be an exact copy of each other. So if I understand correctly, I can just pick any of the 4 drives and back that up to Backblaze B2, correct? So I would not even need to use cloud drive. I think that would be the perfect solution. Thanks for your help!

[Shane]

You're welcome! If the backup software does not do so already, you may wish to exclude the hidden system "$RECYCLE.BIN" and "System Volume Information" folders from your backup to save a bit of space there. If you're not using reparse points (symlinks, junctions, etc) inside the pool you could also exclude the ".covefs" system folder.