RAIDZ is ZFS's software parity — like RAID 5/6 but with checksums, self-healing and no write hole. Here's what makes it different, then size a pool in the ZFS calculator.
RAIDZ vs traditional RAID
RAIDZ1, RAIDZ2 and RAIDZ3 are single, double and triple parity — broadly the ZFS equivalents of RAID 5, 6 and triple-parity RAID. The difference is that ZFS is a combined filesystem and volume manager, so it knows what every block contains. It checksums all data and metadata, detects silent corruption, and self-heals from parity or another copy during a scrub.
ZFS also uses variable-width stripes and copy-on-write, which eliminates the RAID 'write hole' (the corruption risk when power is lost mid-write in traditional parity RAID). That integrity story is why TrueNAS and many NAS platforms are built on ZFS.
vdevs are the key concept
A ZFS pool is made of one or more vdevs (virtual devices); a RAIDZ vdev is a parity group. You add capacity and performance by adding vdevs. Critically, redundancy is per vdev — losing a whole vdev loses the pool — and a RAIDZ vdev delivers roughly the random IOPS of a single drive, so pool IOPS scale with the number of vdevs, not drive count. See RAIDZ1 vs Z2 vs Z3.
For lots of random IOPS (databases, VMs), use more (narrower) vdevs or mirror vdevs; for capacity, use wider RAIDZ2 vdevs.
Real capacity is less than the simple sum
ZFS usable capacity is parity removed, then a ~3.2% slop reservation, then (parity+1)-sector allocation padding — negligible at the default 128 KiB recordsize but significant for small volblocksize (e.g. 8 KiB zvols for VMs). A naïve calculator overstates ZFS capacity; ours models the lot. Keep pools below ~80% full for performance.
Try the ZFS calculator and change recordsize to see padding appear.