ZFS RAIDZ and a hardware RAID controller solve the same problem two very different ways. Here's how they compare on integrity, performance, flexibility and cost. Size either in the RAID and ZFS calculators.
Data integrity — ZFS's big win
ZFS checksums every block and self-heals silent corruption (bit rot) from parity or another copy during scrubs; it also eliminates the parity 'write hole' via copy-on-write. A traditional hardware RAID controller trusts the drives and has no end-to-end checksum, so silent corruption can pass undetected. For data integrity, ZFS is materially stronger — which is why it underpins TrueNAS and many NAS platforms.
Hardware RAID's counter is simplicity and offload: a battery/flash-backed controller handles parity in dedicated silicon, with a cache that absorbs writes, and presents a simple volume to any OS.
Performance and flexibility
Hardware RAID's write cache (BBWC/FBWC) can smooth bursty writes and mask the parity penalty; ZFS uses host RAM (ARC) and an optional SLOG/L2ARC instead. For random IOPS, remember a RAIDZ vdev ≈ one drive — so ZFS scales IOPS with vdevs (or mirrors), while a hardware RAID 10 may give better out-of-the-box random write IOPS on the same disks.
ZFS wins on flexibility: snapshots, compression, replication and clones are built in. Hardware RAID needs the OS/filesystem (or array) to add those. For HBA-passthrough/software-defined storage, see hardware RAID vs HBA passthrough.
Which to choose
Choose ZFS RAIDZ for NAS, backup targets, archives and anywhere data integrity, snapshots and replication matter — give it an HBA (not a RAID card) and ECC RAM. Choose hardware RAID for simple boot volumes, OS-agnostic arrays, or appliances where a controller with cache and a familiar management stack is wanted.
Capacity-wise, ZFS usable is parity + slop + padding (model it in the ZFS calculator), while hardware RAID is the clean parity figure — our RAID calculator covers both.