Glossary

ZFS's in-RAM read cache. It uses most free memory by design; high RAM 'usage' on a healthy TrueNAS box is usually just ARC doing its job.

See also: L2ARC, Recordsize, ECC RAM

A hash ZFS stores for every block to detect corruption on read. Combined with redundancy, it lets ZFS self-heal bad data instead of silently returning it.

See also: Scrub, ECC RAM

Transparent dataset compression. LZ4 is the near-free default; ZSTD trades some CPU for higher ratios. Usually a net performance win because it reduces disk I/O.

See also: Recordsize, Dataset

ZFS never overwrites live data in place; it writes new blocks and updates pointers atomically. This is what makes snapshots cheap and power-loss corruption rare.

See also: ZFS, Snapshot

A named filesystem within a pool with its own properties (recordsize, compression, quotas, snapshots). The unit of organization and tuning in ZFS.

See also: zvol, Recordsize, Snapshot

Memory that detects and corrects single-bit errors. Strongly recommended for ZFS so corruption can't enter via RAM before checksums are computed — important, though ZFS still functions without it.

See also: Checksum, ARC (Adaptive Replacement Cache)

A storage controller flashed to IT mode that passes disks directly to ZFS with no hardware RAID in between. The correct way to attach many drives to TrueNAS.

See also: ZFS, Pool (zpool)

An optional SSD second-level read cache that extends ARC. Only helps specific large-working-set workloads and consumes RAM to index — rarely worth it for home media.

See also: ARC (Adaptive Replacement Cache), SLOG (separate ZIL device)

A vdev where every disk holds the same data. Fast random I/O and quick resilvers, at the cost of 50% usable capacity for a two-way mirror. Often preferred for small home pools.

See also: RAIDZ (Z1/Z2/Z3), vdev (virtual device), Resilver

The top-level ZFS storage unit, built from one or more vdevs. Capacity and redundancy are properties of the vdevs inside it; the pool stripes data across them.

See also: vdev (virtual device), RAIDZ (Z1/Z2/Z3), Mirror

ZFS parity redundancy. RAIDZ1 tolerates one disk failure per vdev, Z2 two, Z3 three. Space-efficient for large vdevs but slower to resilver than mirrors.

See also: Mirror, vdev (virtual device), Resilver

The maximum logical block size for a dataset. Large (1M) suits sequential media storage; smaller values suit databases and random-write workloads. A key tuning lever.

See also: Dataset, ARC (Adaptive Replacement Cache)

Sending snapshots to another pool or system via zfs send/receive for off-box or off-site backup. Incremental after the first transfer, preserving snapshot history.

See also: Snapshot, zfs send / receive

Rebuilding redundancy onto a replacement disk. Mirrors resilver fast; wide RAIDZ vdevs can take long enough that a second failure during resilver is a real risk.

See also: RAIDZ (Z1/Z2/Z3), Mirror, Scrub

A scheduled full read of the pool that verifies every block against its checksum and repairs corruption from redundancy. Should run regularly to catch silent bit rot.

See also: Resilver, Checksum

A fast, power-loss-protected SSD that accelerates synchronous writes by offloading the ZFS Intent Log. It does nothing for async writes and is unnecessary for typical SMB file shares.

See also: ZIL (ZFS Intent Log), L2ARC

A network file share using the SMB/CIFS protocol — the standard way Windows, macOS, and Linux clients mount TrueNAS storage on a LAN.

See also: Dataset

A read-only, point-in-time image of a dataset. Near-instant and space-efficient thanks to copy-on-write — the foundation of TrueNAS data protection and ransomware recovery.

See also: Copy-on-write (CoW), Replication

The FreeBSD-based edition (formerly FreeNAS), historically very stable for pure file serving. Now in maintenance mode as development consolidates on SCALE.

See also: TrueNAS SCALE, ZFS

The Linux-based (Debian) edition of TrueNAS, with native container/app and virtualization support. The recommended default for most new home builds.

See also: TrueNAS CORE, ZFS

A redundancy group within a pool — a mirror, a RAIDZ set, or a single disk. A pool's fault tolerance is per-vdev: losing any full vdev loses the whole pool.

See also: Pool (zpool), RAIDZ (Z1/Z2/Z3), Mirror

The copy-on-write filesystem and volume manager TrueNAS is built on. Combines pooled storage, checksummed integrity, snapshots, and compression in one layer.

See also: Pool (zpool), vdev (virtual device), Copy-on-write (CoW)

The mechanism replication uses: serializing a snapshot (or the delta between two) into a stream that another pool can ingest exactly.

See also: Replication, Snapshot

The on-pool log that guarantees synchronous writes survive a crash. It exists by default; a SLOG only relocates it to faster media.

See also: SLOG (separate ZIL device)

A block device carved from a pool (instead of a filesystem dataset), typically used for iSCSI targets or VM disks.

See also: Dataset