ZFS Storage

ZFS — the safest filesystem

ZFS combines filesystem, volume manager and RAID in one layer. Enterprise quality, available for free.

Architecture

How ZFS works

Pool

Storage Pool (zpool)

Multiple disks form a pool. Redundancy via Mirror or RAIDZ1/2/3. Expandable online.

Data

Dataset

Dynamic filesystems in the pool. No fixed partition size. Quotas and reservations possible.

Zvol

Dynamic block devices (virtual disks) in the pool. No fixed size. Used for iSCSI, VMs and virtual drives.

Snap

Snapshots

Instant read-only copies. No initial storage use. Basis for replication and versioning.

CoW

Copy-on-Write

Data is never overwritten. Guarantees consistency even during power failure.

SHA

Checksums

Every block verified. Errors detected and automatically repaired with redundancy.

Quot

Quota

Size of a ZFS dataset grows dynamically up to pool size. Limit usable size with a quota and guarantee minimum space with reservations.

Res

Reservation

A reservation guarantees a minimum capacity in the pool for a dataset or user. The reserved space is subtracted from free pool space and is not available to other datasets.

ARC

Cache layers

ARC (RAM read cache), RAM write cache with cache protection (SLOG/ZIL), L2ARC (SSD read cache), ZIL/SLOG (sync writes). Intelligent caching for high performance.

RAID

Redundancy options

Type	Min. disks	Fault tolerance	Capacity	Use case
Stripe	1+	none	100%	Cache, temp data
Mirror	2+	1+ disk	max 50%	Boot pool, small pools
RAIDZ1	2+	1 disk	~50-80%	General use, expandable
RAIDZ2	3+	2 disks	~50-75%	Recommended for production, expandable
RAIDZ3	4+	3 disks	~40-70%	Very large pools, expandable
DRAID	3+	configurable	variable	Shortest rebuild time for very large disk counts
AnyRAID	announced		maximum	Any disk sizes, expandable, shrinkable

Recommendation: RAIDZ2 protects against 2 simultaneous disk failures without excessive capacity loss.

Data security

Protection at all levels

Silent corruption protection

ZFS checksums detect silent data errors (bit rot) that other RAID systems miss. Automatic repair with redundancy.

Native encryption

AES-256-GCM directly in the filesystem. Central keyserver possible. Transparent for SMB, NFS, S3.

Snapshot protection

ZFS snapshots are read-only and immutable. Ransomware protection without root access.

Online scrubbing

Regular verification of all data while running. Errors repaired immediately.

Crashproof

Copy-on-Write protects the ZFS filesystem during a crash while writing. No more long fsck or chkdsk disk checks after a system crash.

VM-safe

Sync Write protects confirmed writes in ZIL/SLOG so they can be completed on the next reboot. Virtual machines and databases benefit directly.

Sync Write logging

Protects confirmed writes in ZIL/SLOG after a crash during write. Missing writes are completed on next reboot.

RAID

Protects your data from bad disks. Use a hot spare for immediate rebuilds without waiting for manual intervention.