How Many Drives Can Fail in RAID 5?

The short answer

A RAID 5 array can tolerate exactly one failed drive without losing array availability.

If a second drive fails before the first one is replaced and the array successfully rebuilds, RAID 5 loses its redundancy and may become inaccessible or suffer data loss. Some recovery may be possible depending on the failure and implementation, but it is not something to trust with important data.

RAID 5 is therefore a one-drive-failure design, not a two-drive-failure design and not a backup system.

How RAID 5 protects data

RAID 5 stripes data and parity across at least three drives. The parity information allows the array to reconstruct the missing contents of one failed drive.

The basic capacity formula is:

Usable capacity ≈ (number of drives - 1) × capacity of the smallest drive

The smallest drive sets the usable size because RAID generally uses an equal amount of space from each member drive. Actual usable space will be lower after filesystem, metadata, and NAS or RAID-controller overhead. Drive manufacturers also advertise decimal capacity, while some operating systems display capacity in tebibytes (TiB).

What happens after one drive fails?

The array enters a degraded state:

• Data may remain available.
• The array has no remaining drive-failure tolerance.
• Reads for missing data must be reconstructed from the surviving drives and parity.
• Performance can change during degradation; the exact impact depends on the NAS, RAID implementation, workload, and drive condition.
• A replacement drive must be installed and the array must complete its rebuild before full redundancy returns.

A second drive failure during this period is the critical risk. A rebuild can also expose latent read errors or other problems on the surviving drives, which is one reason RAID 5 becomes less attractive as arrays grow larger or contain more valuable data.

Worked RAID 5 example

Suppose a NAS uses four 12 TB drives in RAID 5:

(4 - 1) × 12 TB = 36 TB theoretical usable capacity

This array can tolerate:

• One failed drive: yes, assuming the array rebuilds successfully.
• Two failed drives at the same time: no.
• A second failure during rebuild: no reliable protection.

The 36 TB figure is before filesystem and system overhead. The capacity shown by the NAS may appear lower, and the result can differ if the installed drives are not exactly equal in usable capacity.

If one 12 TB drive fails, the array still presents its data while operating in degraded mode. After replacing that drive, the NAS reconstructs its contents from the surviving data and parity. Until the rebuild completes, another drive failure can put the array at serious risk.

Rebuild behavior and risk

A RAID 5 rebuild is more than a simple file copy. The NAS must reconstruct the missing member across the array, often while handling normal reads and writes from users, applications, Plex, surveillance recording, or backup jobs.

Rebuild duration is not a fixed number. It depends on factors such as:

• Drive capacity and condition
• RAID software or controller behavior
• Number of drives and the array layout
• Current NAS workload
• Network and storage activity
• Whether the NAS prioritizes rebuild speed or normal services
• Errors encountered on surviving drives

During a rebuild, avoid unnecessary intensive workloads where practical. A NAS recording security cameras, serving multiple Plex streams, synchronizing cloud data, or running another large backup may take longer to rebuild than an idle system.

A UPS can help protect the NAS from abrupt power loss during normal operation and rebuilds, but it does not increase RAID 5's failure tolerance. Use a UPS with compatible NAS shutdown support and test the shutdown behavior.

RAID 5 capacity versus protection

RAID 5 uses the equivalent of one drive for parity. This is more space-efficient than mirroring, but its protection is limited.

RAID level	Minimum drives	Drive failures tolerated	Approximate usable capacity	Main trade-off
RAID 5	3	1	(N - 1) × smallest drive	Efficient capacity, but only one-drive protection
RAID 6	4	2	(N - 2) × smallest drive	More protection, less usable capacity
RAID 10	4	Depends on which drives fail	About 50% of raw capacity	Strong rebuild profile, but less capacity efficiency
Single-disk or JBOD	1 or more	None	Depends on layout	No array redundancy

RAID 10 can survive more than one failure if the failed drives are in different mirrored pairs. If both drives in the same mirror fail, that mirror is lost. Its tolerance is therefore not expressed as a single guaranteed number in the same way as RAID 5 or RAID 6.

Capacity calculations are estimates. A specific NAS may reserve space, use a different RAID layout, or impose expansion rules that change the result.

Expansion flexibility

RAID 5 may be expandable, but expansion is implementation-dependent. A NAS might support one or more of these approaches:

1. Replace each drive with a larger drive, rebuilding after each replacement.
2. Add drives to an existing RAID group.
3. Create a separate storage pool or volume.
4. Migrate data to a new, larger array.

Not every NAS supports every method. Some expansions require all existing drives to be replaced before the additional capacity becomes available. Others require an online reshape that can take a long time and temporarily increase storage stress.

Before buying, check the exact NAS documentation for:

• Supported RAID migration paths
• Whether drives can be added to an existing RAID 5 group
• Whether all drives must be the same size
• Whether expansion is online or requires downtime
• Minimum and maximum array or pool sizes
• Whether expansion changes the filesystem or volume layout
• What happens if an expansion is interrupted

ZFS deserves separate caution. A RAIDZ1 vdev provides single-drive parity protection similar in concept to RAID 5, but vdev expansion and pool growth depend on the ZFS implementation and version. Do not assume that a ZFS NAS expands the same way as a hardware RAID or a conventional NAS RAID group.

RAID is not backup

RAID 5 protects availability after one drive failure. It does not protect against:

• Accidental deletion
• Ransomware or malware
• File corruption replicated across the array
• Theft, fire, flood, or other physical damage
• NAS theft
• A failed RAID controller or enclosure
• Incorrect permissions or synchronization
• Multiple drive failures
• A damaged filesystem or failed rebuild

A practical backup plan should include copies outside the RAID 5 array. For important files, use the 3-2-1 principle:

• Keep at least three copies of the data.
• Use at least two different types of storage or media.
• Keep at least one copy off-site or otherwise separated from the NAS.

A second NAS, external drive rotated away from the main system, or encrypted cloud backup can serve as an additional destination. Test restores rather than assuming that a completed backup job proves the files are recoverable.

An SSD cache does not change RAID 5's failure tolerance either. Cache can affect application behavior and workload handling, but it is not a substitute for a backup or a second parity drive. For Plex, RAID mainly provides storage availability for the media library; it does not protect the original media or application data from deletion. For surveillance, RAID may keep recordings available after one drive fails, but retention settings, camera bitrate, and a separate export or backup plan still matter.

When RAID 5 makes sense

RAID 5 can be reasonable when:

• The NAS has at least three drive bays.
• Usable capacity matters more than dual-drive fault tolerance.
• The data is also backed up elsewhere.
• The workload and array size are appropriate for the NAS implementation.
• You can monitor drive health and replace failed drives promptly.
• You understand the rebuild and expansion process before deployment.

It is less suitable when:

• The data cannot be recreated or restored quickly.
• The array will contain many large drives.
• A long degraded or rebuild period is unacceptable.
• The NAS will hold the only copy of important files.
• You need guaranteed tolerance for any two drive failures.
• The cost of losing availability is greater than the capacity saved over RAID 6 or RAID 10.

Before selecting a NAS, compare the available drive bays, supported RAID levels, expansion method, backup targets, and networking needs. You can Browse NAS or review NAS and storage servers while checking those constraints.

RAID decision table

Your priority	More suitable starting point	Why
Maximum usable capacity with one-drive protection	RAID 5	Sacrifices the equivalent of one drive for parity
Protection from any two drive failures	RAID 6	Uses dual parity and tolerates two failed drives
Stronger rebuild behavior and predictable mirroring	RAID 10	Mirrors data and stripes across mirror pairs
A small home media library with replaceable content	RAID 5 may be adequate	Capacity efficiency can be useful if the library is backed up or reproducible
Irreplaceable photos, documents, or business data	RAID 6, RAID 10, or another suitable design plus backups	RAID level alone cannot provide data protection
ZFS-based storage	RAIDZ level selected for the workload	Check vdev layout, expansion, and recovery behavior for the specific implementation
Surveillance storage	Choose based on retention, drive bays, and backup needs	RAID does not determine camera capacity or replace off-system retention
Plex storage and streaming	Choose based on capacity, network, and transcoding needs	RAID protects availability, not media ownership or application recovery

The exact answer remains simple: RAID 5 can lose one drive. It cannot safely tolerate two failed drives. If the data matters, pair RAID with monitoring, a tested replacement plan, and independent backups.