Creating a storage pool (referenced briefly in the earlier TrueNAS Scale introduction) asks you to choose a redundancy layout upfront — a decision that’s genuinely difficult to change later without rebuilding the pool from scratch. Understanding ZFS pool layout options properly before creating your pool saves you from discovering the wrong tradeoff months into using your NAS.

Why This Decision Matters So Much

Unlike many storage decisions that can be adjusted later, changing a ZFS pool’s redundancy layout (going from a mirror to RAID-Z2, for instance) generally requires creating an entirely new pool and migrating data across, rather than a simple reconfiguration — making this genuinely one of the more consequential upfront decisions in setting up a TrueNAS system.

Mirror: Simplicity and Performance

A mirror layout duplicates data identically across two (or more) drives, similar in concept to traditional RAID 1. Every drive in the mirror holds a complete copy of the data.

Pros: fastest read performance among redundancy options, simplest to understand and expand (add another mirrored pair), fastest resilver (rebuild) time if a drive fails Cons: the least storage-efficient option — a two-drive mirror gives you the usable capacity of just one drive, meaning 50% of raw capacity is dedicated to redundancy

Best for: performance-sensitive workloads (a pool serving as Proxmox VM storage via NFS, per the earlier shares guide), or setups prioritizing fast rebuild times and simplicity over maximum storage efficiency

RAID-Z1: Balanced Efficiency and Redundancy

RAID-Z1 is ZFS’s equivalent to traditional RAID 5, striping data with a single parity block across the array, tolerating exactly one drive failure without data loss.

Pros: better storage efficiency than mirroring — a 4-drive RAID-Z1 array gives you 3 drives’ worth of usable capacity Cons: during a drive failure and subsequent resilver, the array operates in a degraded state with zero further fault tolerance — a second drive failure during that resilver window means data loss; resilver time also grows considerably with larger drives, extending this vulnerable window

Best for: homelab users comfortable with the single-fault tolerance tradeoff, prioritizing storage efficiency over the fastest possible rebuild safety margin, generally most reasonable with smaller-capacity drives where resilver time stays manageable

RAID-Z2: Stronger Redundancy for Larger Arrays

RAID-Z2 uses double parity, tolerating two simultaneous drive failures — genuinely important as drive capacities have grown, since larger drives mean longer resilver times, extending the vulnerable single-fault-tolerance window RAID-Z1 exposes you to.

Pros: tolerates two drive failures, considerably safer for larger drive capacities and arrays where resilver time is genuinely extended Cons: more parity overhead than RAID-Z1 — a 6-drive RAID-Z2 array gives you 4 drives’ worth of usable capacity, a less efficient ratio than the equivalent RAID-Z1 configuration

Best for: larger capacity drives (the resilver-time concern becomes genuinely relevant beyond a few terabytes per drive), or genuinely important data where you want protection against a second failure occurring during an already-degraded rebuild

A Practical Sizing Framework

• 2 drives, prioritizing simplicity/performance → Mirror
• 4-6 smaller drives, cost-conscious → RAID-Z1
• 6+ drives, or larger-capacity drives specifically → RAID-Z2
• Genuinely mission-critical data with maximum redundancy tolerance → RAID-Z3 (triple parity), though this is generally overkill for homelab scale and rarely necessary outside genuine enterprise requirements

Why Drive Count Also Matters

ZFS’s parity calculations perform most efficiently with specific drive count patterns (often recommended as 2^n + parity drives, e.g., 4+1 for RAID-Z1, 4+2 for RAID-Z2), though modern ZFS handles non-optimal counts reasonably well for typical homelab scale — this optimization matters more at genuine enterprise scale than for most homelab pool sizes.

Multiple Mirrors as an Alternative to Wide RAID-Z

Rather than one wide RAID-Z2 array, some homelab users configure multiple mirrored pairs combined into a single pool (similar in spirit to RAID 10), trading some storage efficiency for better performance and simpler individual-pair rebuild characteristics — worth considering specifically for the same VM storage performance priorities that favor mirrors generally.

Combining Layout Choice with ECC RAM

Given the ECC RAM guide covered earlier specifically discussed ZFS’s data integrity promises depending on uncorrupted RAM, the redundancy layout you choose here works alongside, not instead of, that consideration — RAID-Z2’s drive-failure protection doesn’t substitute for the bit-flip protection ECC specifically addresses; they protect against genuinely different failure modes.

Final Thoughts

Choosing between mirror, RAID-Z1, and RAID-Z2 for your ZFS pool layout means weighing storage efficiency against redundancy strength and rebuild safety margins, a decision worth genuine consideration given how difficult it is to change after the fact. For most homelab NAS builds using reasonably modern, moderate-capacity drives, RAID-Z2 offers a sensible default balance — but understanding when a mirror’s performance or RAID-Z1’s efficiency genuinely fits your specific situation better helps you make a deliberate choice rather than defaulting to whichever option a general guide happened to recommend.