[s26] Why the Store relies on ZFS

The Store relies on ZFS not just for snapshotting, compression and other very convenient features, but mainly because of reliability.

Copy-on-Write filesystems like ZFS make sure one can never see corrupt (due to partial overwrite) blocks. Data corruption is quite frequent in multi-terabyte storage, and it usually happens silently.

That is why ZFS is so important to the store.

Other positive aspects for a sysadmin to use ZFS

  • He can take snapshots and revert complete datasets
  • He can do boot environments so updating is safe
  • He can trust that data doesn’t get corrupted by defective hardware (i.e. disk controller)
  • He can run periodic scrubs and automatically correct errors
  • It is easy to do a ZFS send/receive to perform a backup

Encryption possibility

FreeBSD’s FDE (Full Disk Encryption) is called geli and it works with any filesystem, it just provides an encrypted disk device. ZFS is going to have native encryption soon, too.


  • SCSI (SAS, USB) disks are /dev/da#
  • ATA (SATA) disks are /dev/ada#.

da means Direct Access. There is “indirect access” — e.g. LSI MegaRAID cards will provide /dev/mfid# disks.

Partitions are /dev/ada#p#.

Do not use these # numbers though. There is a better way — /dev/gptid/* provides access by GPT UUID, /dev/gpt/* provides access by GPT label, and /dev/diskid/* provides access by GPT disk ID.

[[[/s26/][[s26] Why the Store relies on ZFS]], last update 2017-12-08]