RAID 5 vs RAID 6 for a NAS: Capacity, Rebuild Risk and When Each Makes Sense
Quick answer: RAID 5 uses one drive-equivalent of capacity for distributed parity and tolerates one drive failure. RAID 6 uses two drive-equivalents and tolerates two failures. On four 12TB disks, that means 36TB nominal RAID-usable capacity with RAID 5 versus 24TB with RAID 6, before NAS and filesystem overhead. RAID 6 provides an extra margin during degraded operation and rebuilding; RAID 5 leaves more room for data. The better choice depends on how many drives you have, what losing access would cost, and whether you can restore reliably from a separate backup.
The difficult question is not which RAID level sounds safer. It is whether the additional drive-equivalent consumed by RAID 6 is worth the added failure tolerance for your workload. This guide evaluates that choice for conventional NAS hard-drive arrays, using documented RAID behavior and worked configurations rather than lab benchmarks or predictions about when drives will fail.
RAID 5 vs RAID 6 at a glance
| Question | RAID 5 | RAID 6 |
|---|---|---|
| Minimum active drives | 3 | 4 |
| Parity overhead | One drive-equivalent, distributed across members | Two drive-equivalents, distributed across members |
| Nominal capacity with equal-sized disks | (N − 1) × disk size | (N − 2) × disk size |
| Whole-drive failures tolerated | One | Two |
| After one failed disk | No further whole-drive failure tolerance | Can still tolerate one more whole-drive failure |
| Typical reason to consider it | More usable capacity from limited bays | Extra protection during degraded operation and recovery |
| Does it replace backup? | No | No |
Neither layout keeps all parity on a designated disk. Parity is distributed across members, while the formulas describe the capacity equivalent allocated to protection. These are planning estimates for equal-size active disks. The storage pool and file-available capacity can be lower. For the complete TB-to-TiB and filesystem explanation, see how to calculate usable NAS storage.
What does RAID 6’s extra parity actually buy?
Imagine a four-disk RAID 5 array with one disk failed. The NAS may remain online, but the array is degraded. If another member disk becomes unavailable before the first is repaired, the array no longer has the complete information required to reconstruct its data. RAID 6 starts with a second independent parity relationship: after the first member fails, it normally retains the ability to tolerate a second member failure.
That extra protection is valuable during the interval between failure detection and completed rebuilding. It is not a guarantee against controller failure, data deletion, malicious changes, filesystem corruption, or loss of the entire NAS. Nor does it make the array immune to every unreadable block or implementation failure.
RAID 5 vs RAID 6 capacity with 4, 6 and 8 drives
Use equal-sized active disks so the comparison is meaningful. The following figures are nominal decimal terabytes before platform-specific deductions.
| Configuration | Raw TB | RAID 5 TB | RAID 6 TB | Additional RAID 6 overhead |
|---|---|---|---|---|
| 4 × 12TB | 48 | 36 | 24 | 12TB |
| 4 × 20TB | 80 | 60 | 40 | 20TB |
| 6 × 12TB | 72 | 60 | 48 | 12TB |
| 6 × 20TB | 120 | 100 | 80 | 20TB |
| 8 × 20TB | 160 | 140 | 120 | 20TB |
RAID 6’s additional parity always costs one more drive-equivalent than RAID 5 in these standard equal-size examples. But the percentage sacrifice falls as the array grows. RAID 6 is 50% capacity-efficient with four drives, about 66.7% with six, and 75% with eight. RAID 5 is 75%, about 83.3%, and 87.5% respectively.
Want to change the disk count or size? Use the RAID & NAS Capacity Calculator rather than assuming four-bay results scale directly to larger enclosures. The 2-bay vs 4-bay NAS comparison explains why drive bays themselves are part of the storage decision.
The four-bay trade-off: what you give up for dual parity
Four bays are where the choice feels most expensive. Four 20TB disks cost the same to purchase whether you configure RAID 5 or RAID 6, but the nominal protected capacity changes from 60TB to 40TB. If 40TB will not cover your data and realistic growth, RAID 6 does not solve the sizing problem merely because it has more parity. You may need larger disks, more bays, or a different storage architecture.
Consider a video editor with 25TB already stored and another 5TB expected each year. A four-drive 12TB RAID 5 array offers 36TB nominal RAID capacity before overhead; RAID 6 offers 24TB. Neither should be selected by looking at drive labels alone. The editor must establish the actual file-space target, required working headroom, backup scheme, and cost of restoring the archive. The NAS capacity-planning guide is the appropriate starting point.
Compare drive cost per usable TB, not just raw TB
Take four identical disks costing P dollars each. Total drive cost is 4P. With 20TB drives, RAID 5 yields 60TB nominal and RAID 6 yields 40TB. The drive cost per nominal usable TB is therefore 4P/60 versus 4P/40, making RAID 6’s figure 50% higher for that four-disk example. This is a relative calculation, not a claim about current disk prices or total cost of ownership. It excludes enclosure, electricity, replacements, backups, and other expenses.
At eight equal drives, RAID 5 provides seven drive-equivalents while RAID 6 provides six: the relative drive-cost-per-usable-TB penalty shrinks to about 16.7%. If dual parity is important, selecting a larger-bay NAS can change the economics materially. It can also raise hardware and operating costs, so compare the whole system rather than capacity efficiency alone.
What happens when a drive fails?
1. Detection: The NAS reports a disk or array fault. Alerts matter because an unnoticed degraded array extends the period of exposure. Confirm the failed disk’s identity in the storage manager before touching hardware.
2. Degraded operation: The NAS may continue serving data. RAID 5 has exhausted its whole-drive failure tolerance; RAID 6, after the first failure, ordinarily retains one further failure’s tolerance. Do not mistake a still-accessible shared folder for a healthy array.
3. Replacement or spare activation: A compatible replacement must be installed, or a configured hot spare may be activated. Follow the specific NAS vendor’s drive replacement process. Do not pull an additional disk simply because the array is already degraded.
4. Rebuild/resync: The array reconstructs missing information. Operations can take significant time and compete with normal traffic for disk and controller resources. The system may remain vulnerable until redundancy is restored.
5. Verification: Confirm a healthy array state, resolve any warnings, check the backup job, and investigate why the failure occurred. An array returning to “Ready” does not independently prove that every protected file is recoverable from backup.
Synology publishes operation-time estimates with explicit assumptions and notes that other running services, drive condition and resync priorities alter real-world durations. QNAP offers “Service First” and “Resync First” modes precisely because rebuild speed and foreground responsiveness compete. Thus a prediction such as “a 20TB RAID 5 rebuild takes exactly 18 hours” cannot be justified from drive size alone.
UREs, large hard drives and the RAID 5 rebuild argument
An unrecoverable read error (URE) occurs when a storage device cannot return requested data even after its internal error-correction process. During a degraded RAID 5 rebuild, the system may need data from remaining members to reconstruct missing blocks. If necessary data cannot be read and no sufficient redundancy remains, recovery of the affected information may fail. The result depends on what was unreadable and the RAID implementation; it is not accurate to insist that every URE destroys an entire array.
Manufacturers publish nonrecoverable-read-error specifications, often expressed as a rate such as one error per 1014 or 1015 bits read, depending on the exact drive. These specifications are inputs to reliability analysis, not a timer that guarantees a failure when an array reads a particular number of terabytes. Applying a simple probability formula assumes a particular error model, independence, data-read volume and recovery behavior. Real rebuild outcomes also depend on workload, previously discovered bad sectors, media condition and controller/software behavior.
An IBM analysis of RAID 5 and RAID 6 shows why unreadable data and longer recovery windows deserve consideration, while also relying on modeled assumptions. It does not establish that RAID 5 becomes categorically unusable at a universal 12TB, 16TB or 20TB threshold.
For larger disks, ask better questions: How long would this particular platform take to rebuild under your workload? Are drive warnings and parity checks monitored? Can the data be restored independently? Can work continue during the restore? RAID 6’s extra parity offers a meaningful margin, but neither RAID level removes the need for those answers.
Is RAID 6 slower than RAID 5?
Both arrays read and write across multiple disks, and RAID 6 must maintain more parity information. Small, partial-stripe writes can require additional reads and parity calculations; full-stripe sequential writes behave differently. Frequently quoted “write penalties” are simplified I/O models, not universal speed measurements for any NAS.
CPU capability, disk speed, queue depth, file size, network ports, caching, filesystem settings and simultaneous users can matter as much as RAID type. A home NAS connected through 1GbE may be constrained by the network in a workload where array-level differences are hard to observe. The opposite may be true for a high-throughput multi-user editing system. Without tests on the specific hardware, an honest comparison should not promise particular MB/s figures or declare RAID 6 unsuitable for Plex or backup use.
RAID 5 plus a hot spare is not the same as RAID 6
A hot spare is an idle configured drive that can be brought into service automatically when a RAID member fails. QNAP’s hot-spare documentation confirms that the spare normally does not store active data. Once a member fails, the NAS starts rebuilding onto the spare.
Take five 12TB drives: with four active disks in RAID 5 and one hot spare, nominal usable capacity remains 36TB. With five active disks in RAID 6, nominal usable capacity is also 36TB. The protection mechanism differs: RAID 6 has two active parity drive-equivalents from the outset; RAID 5 plus a hot spare begins repairing automatically but has no second level of active parity before reconstruction restores redundancy. Whether to reserve a bay for a spare or use it for parity depends on platform support and recovery priorities.
Can you convert RAID 5 to RAID 6 later?
Sometimes, but never assume a full four-bay RAID 5 array can be converted in place without preparation. Supported migration paths typically require an additional free, compatible drive. Both QNAP QTS 5.2 documentation and Synology DSM documentation describe RAID 5-to-RAID 6 migration with at least one additional drive for supported configurations. Some NAS models or storage layouts do not support it at all.
If all bays are occupied, you may have no straightforward supported online migration route. The practical alternative can involve verified backup, replacing or reconfiguring hardware, destroying/recreating the pool, and restoring data. Such operations can be time-consuming and carry their own risks. A vendor’s migration feature is not a substitute for a verified backup.
Also distinguish conventional RAID from Synology SHR/SHR-2 and ZFS RAIDZ1/RAIDZ2. Their capacity allocation, upgrade pathways and implementation details can differ. Confirm the exact storage manager and model before following a generic RAID tutorial.
How to reduce avoidable storage risks whichever RAID you choose
- Enable alerts: make sure failed-disk and degraded-pool notifications reach someone who will act.
- Check drive health: review SMART and vendor diagnostic warnings, rather than waiting for outright failure.
- Schedule supported checks: use RAID parity scrubbing/consistency checks as recommended by your platform. These can reveal certain latent problems and may temporarily reduce performance; they do not guarantee future recovery.
- Maintain a tested backup: at least one independent copy should be restorable without relying on the same NAS and disks.
- Plan replacement logistics: know which drive types are approved, how to identify slots, and how quickly you can obtain replacements.
- Protect power: where appropriate, use a properly configured UPS and graceful shutdown so an interruption does not compound an already degraded situation.
QNAP provides scheduled RAID scrubbing and resync-priority controls. The right schedule and procedure depend on the NAS software, array workload and documentation. Maintenance lowers some risks; it cannot turn either RAID level into a backup system.
Which layout fits your situation?
| Situation | What to investigate first |
|---|---|
| Three-disk home NAS | RAID 5 if conventional parity RAID is the goal; RAID 6 needs at least four active disks. |
| Four-bay NAS, limited space and good backups | Whether RAID 5’s extra drive-equivalent makes the system properly sized; understand recovery downtime. |
| Four-bay NAS, critical work and costly downtime | Whether RAID 6’s capacity penalty is acceptable or a larger-bay device is justified. |
| Six- or eight-bay NAS with large HDDs | RAID 6’s smaller proportional capacity penalty and extra degraded-state tolerance. |
| Large replaceable media collection | Restore time and backup/source availability may weigh heavily alongside capacity. |
| Business or creative files that cannot be easily reconstructed | Resilience, notifications, support, backups and recoverability; consider RAID 6 as one part of the plan. |
A five-question RAID decision worksheet
- How many bays can you actually populate? Count active bays and any spare you intend to reserve.
- What is your true file-space requirement? Include expected growth and operational headroom rather than today’s stored bytes alone.
- What is your independent backup and restore plan? Has it been tested with representative data?
- How costly is degraded operation or downtime? Consider people, business obligations and application availability.
- Can the platform migrate or expand later? Verify documented limits before committing every bay.
Put both layouts through the RAID calculator. Then examine actual NAS devices by bay count and compare exact hard drive models. A sensible RAID choice requires both the right protection layout and compatible, supportable hardware. Review the manufacturer’s drive compatibility list before purchasing.
Questions NAS owners ask about RAID 5 and RAID 6
Is RAID 5 still a reasonable choice with modern large hard drives?
Yes, it can be, provided the capacity benefit matters, the restore plan is credible, and the user accepts that RAID 5 has no additional whole-drive failure tolerance after one member fails. A universal drive-size cutoff is not supported by the evidence.
Is RAID 6 worth giving up half the capacity of a four-bay NAS?
It may be when another failed drive during degraded operation would impose unacceptable disruption and the reduced capacity still meets your needs. If it does not fit your storage requirement, compare larger disks or a larger enclosure rather than pretending the shortfall disappears.
Does RAID 6 guarantee no data loss during rebuilding?
No. It tolerates two whole-drive failures in a supported conventional configuration, but cannot prevent all corruption, administrative mistakes, controller failures, or site-wide disasters.
Does RAID 5 automatically fail when rebuilding 20TB disks?
No. Large-drive rebuilds warrant careful risk planning, but a URE specification is not proof of inevitable array failure at a fixed capacity. Hardware health, workload, data-read requirements and implementation matter.
Is RAID 5 with a hot spare as protective as RAID 6?
No. A spare can reduce the delay before a rebuild starts, but it does not provide active second-parity protection while it waits idle.
Can a full four-bay RAID 5 NAS switch to RAID 6?
Not necessarily. Supported online conversion frequently requires adding a disk, which may be impossible if all bays are full. Verify your NAS model and software version before relying on a future migration.
Will RAID 6 slow down Plex streaming or normal backups?
It depends on the specific CPU, disks, network and workload. Additional parity work does not establish a universal real-world performance penalty. Compare tested hardware under workloads similar to yours.
If my files are backed up, why choose RAID 6?
Backup makes loss recoverable, whereas extra parity can help maintain availability through a second member failure. RAID 6 may reduce disruption even when backups are excellent. The value depends on restore time, workload and acceptable downtime.
Bottom line
Choose RAID 5 for its capacity advantage when you have a reliable recovery plan and understand the degraded-state trade-off. Give RAID 6 serious consideration when an additional drive failure during recovery would be especially disruptive, and when its capacity cost fits your design. Neither is a substitute for backup. Count the drive bays, calculate nominal capacity, verify the recovery process, and choose hardware around the resulting requirements—not around a slogan that one RAID level is always right.
Technical sources and methodology
This is a specification- and documentation-based comparison of conventional NAS HDD arrays, not a claim of hands-on DiskBarn performance testing. Capacity tables use equal-sized drives and standard RAID formulas; they exclude filesystem and platform reservations. Rebuild behavior and migration requirements are platform-dependent. Supporting documentation:
- IBM — Re-Evaluating RAID-5 and RAID-6 for Slower, Larger Drives (model-based reliability discussion).
- QNAP QTS 5.2 — Changing RAID type (supported conversions and added disks).
- Synology DSM — Changing the RAID type of a storage pool (requirements and model restrictions).
- QNAP QTS 5.2 — Storage global settings (resync and RAID scrubbing controls).
- QNAP QTS 5.2 — Configuring a RAID group hot spare (spare operation).
- Synology — Estimated time of storage operations (operating assumptions and real-world limitations).
