ZFS is a combined file system and logical volume manager designed for high data integrity. Unlike traditional filesystems, ZFS merges physical storage management and logical volume control into a single stack, eliminating the need for hardware RAID controllers.
How it works
The system operates on the Copy-on-Write (CoW) principle: when data is modified, it is written to a new block rather than overwriting the old one. This ensures filesystem consistency even during power failures. Every data block is verified with a checksum upon reading to detect and automatically repair silent data corruption (bit rot) using redundant copies.
- Snapshots: near-instantaneous point-in-time copies without data duplication.
- ZRAID: a software RAID implementation that solves the classic RAID write hole.
- ARC: an advanced caching algorithm that utilizes system RAM for performance.
ZFS is widely used in enterprise storage and virtualization environments. For example, executing zfs snapshot tank/data@backup creates a recovery point in milliseconds regardless of the dataset size. The architecture theoretically supports pools up to 256 quadrillion zettabytes.