Textbook of Linux

ZFS (originally "Zettabyte File System") is a copy-on-write filesystem and integrated volume manager developed at Sun Microsystems in the early 2000s and released as open source in 2005. It pioneered many features now widely expected: pooled storage across multiple devices, end-to-end checksums, transparent compression, instant snapshots and clones, built-in RAID (RAID-Z), and self-healing when a checksum mismatch is detected on a redundant copy.

On Linux, ZFS is available via OpenZFS, the actively maintained fork that unifies Linux, FreeBSD, illumos, and macOS support. Due to licence incompatibility (CDDL vs GPL), ZFS cannot be merged into the Linux kernel proper, so it is distributed as an out-of-tree module (zfs-dkms). Ubuntu has shipped it as an installer option for years; most other distributions require a manual install.

Core concepts are the zpool (a pool of physical storage composed of vdevs, each a RAID configuration of disks) and the dataset (a filesystem carved out of a pool, with independent compression, quota, and snapshot settings). ZFS's ARC cache, its integration of volume management and filesystem layers, and its operational maturity make it beloved for serious storage. The price is memory: ZFS is hungry for RAM, and deduplication in particular can devour it.