Dip into ZPool

Lately I have been getting the hang of the ZFS filesystem. OpenZFS is great software that combines the functions of disk/volume management and a file system.

By calculating and storing a checksum of your files on the disk alongside the data ZFS can detect file level corruption. ZFS is included with modern linux distros. it can be used for a single disk installation on a workstation or desktop to ensure file integrity. A single drive configuration just scratches the surface of ZFS, the real fun comes when used with multiple disks

If configured in RaidZ, A ZFS volume can survive the loss of couple drives. It accomplishes this by distributing the files and checksum details across a pool of disks to effectively create a software raid array.

In my experiment I have 48x 1TB consumer SSD's loaded into what would now be considered legacy Netapp disk shelves. I formatted the ZPools with RaidZ3, each can survive a 3 disk failure that leaves the file system with ~34 TB of storage.

The clever thing about ZFS is that information about the how the pool of drives is configured is stored alongside the data on every one of the drives. These small partitions contain enough details about the ZPool to allow you to migrate the storage to a new server via Export and Import.

As I work on this further I am looking forward to running some benchmarks and trying various configurations including a single 48 disk pool.