ZFS, Actually Storage you can finally reason about.

The Pool

concept
depth 2/3 − deeper + surface

Every computer you've owned has made you think about which disk things live on: this drive is full, that one has room, move the photos. ZFS retires that question. You hand it all your disks, and it hands back one thing—the pool: a single shared supply of storage the whole machine draws from. Nothing gets carved up in advance, no sizes decided ahead of time, no "resizing" ever again. Everything simply shares the pot.

One warning belongs this early: the pool itself keeps no spare copies of anything. Protection against a dead disk lives one level down, inside the small teams of disks the pool is built from (called vdevs—the next idea to meet). That distinction is the difference between a safe setup and a time bomb, which is why it gets its own page.

The containment hierarchy, once and cleanly:

  • Disks are grouped into vdevs (a mirror pair, a RAIDZ set).
  • Vdevs are striped together into the pool.
  • Datasets carve the pool's one space into policy boundaries.

Day to day, the pool is mostly a health object: zpool status (is it whole), zpool list (how full—watch that number), zpool scrub (verify it). You create it once with zpool create, deliberately, and then live inside it via zfs commands. Homelab reality: most machines want exactly one pool—split pools only when the physics differ (an SSD pool for VMs, an HDD pool for bulk), because separate pools don't share free space.