RAID Planning Checklist Before You Buy Drives

The short answer

Start with your required usable capacity, then add free-space headroom, choose how many drive failures the array must survive, and only then select drive count, RAID level, and spare strategy.

A practical order is:

1. Define the data you need to store today.
2. Add room for growth and free space.
3. Decide whether the array must survive one or two drive failures.
4. Compare usable capacity, rebuild behavior, and expansion options.
5. Decide whether you need a hot spare or a cold spare.
6. Create a separate backup plan.

RAID improves availability. It does not replace backup.

1. Define the capacity target before selecting drives

Do not treat the advertised capacity of all drives as usable NAS capacity. RAID reserves capacity for redundancy, and the filesystem, NAS operating system, snapshots, metadata, and formatting also consume space.

Use this basic estimate:

Nominal RAID capacity = usable drive capacity × data-bearing drive count

For common layouts:

• RAID 5: (number of drives - 1) × smallest drive capacity
• RAID 6: (number of drives - 2) × smallest drive capacity
• RAID 10: number of drives ÷ 2 × drive capacity
• RAID 1: approximately one drive's capacity
• RAID 0: number of drives × drive capacity, with no drive-failure tolerance

These are planning estimates, not guarantees of the free space shown by a NAS. If an array uses drives of different sizes, many RAID implementations base the calculation on the smallest drive.

Add headroom for growth and free space

A capacity target should include:

• Current files
• Expected growth during the array's useful life
• Temporary files and downloads
• Snapshots or version history
• Plex media additions and transcoding-related working space
• Surveillance recordings and retention targets
• A free-space reserve for normal operation and safer maintenance

A simple planning formula is:

Required nominal capacity = target stored data / (1 - desired free-space percentage)

For example, if you expect to store 20 TB and want to keep 20% of the array free:

20 TB / 0.80 = 25 TB

You would then choose an array with at least 25 TB of nominal RAID capacity, before accounting for filesystem overhead and the difference between decimal drive labels and the capacity displayed by the NAS.

2. Decide how many drive failures you must tolerate

Failure tolerance is not the same as the number of drives in the array. It is the number of drive failures the selected layout can survive while remaining accessible.

RAID layout	Typical minimum	Approximate usable capacity	Drive-failure tolerance
RAID 0	2	All drive capacity	None
RAID 1	2	One drive's capacity	One drive
RAID 5	3	(N - 1) × smallest drive	One drive
RAID 6	4	(N - 2) × smallest drive	Any two drives
RAID 10	4	About half of total capacity	One drive per mirror pair, depending on which drives fail
RAIDZ1	Usually 3	Similar to single-parity RAID planning	One drive
RAIDZ2	Usually 4	Similar to dual-parity RAID planning	Two drives

RAIDZ layouts are ZFS storage-pool layouts rather than conventional hardware RAID levels. Exact capacity, expansion, and disk replacement behavior depends on the ZFS implementation and NAS platform.

One failure versus two failures

RAID 5 and RAIDZ1 preserve more capacity than dual-parity layouts, but they provide less failure tolerance. During a rebuild, the array is exposed to another drive failure. A second failure can make a RAID 5 or RAIDZ1 array unavailable or require recovery from backup.

RAID 6 and RAIDZ2 sacrifice additional capacity for protection against two simultaneous or overlapping drive failures. This can be a more conservative choice for larger arrays, important data, or situations where replacing a failed drive may take time.

RAID 10 has a different risk profile. It can survive one failed drive in every mirror pair, but two failures in the same pair can destroy the array. Its usable capacity is usually about half of raw capacity, and it can offer simpler rebuilds because a failed mirror member is copied from its surviving partner.

3. Compare rebuild behavior, not just usable capacity

A rebuild replaces the missing member of a redundant array after a drive failure. Rebuild duration depends on drive capacity, workload, interface, NAS hardware, RAID implementation, and how full the array is. Do not rely on a fixed rebuild-time estimate unless the NAS vendor provides one for the exact configuration.

During a rebuild:

• The array may have reduced performance.
• A RAID 5 or RAIDZ1 array has no remaining single-drive redundancy.
• A RAID 6 or RAIDZ2 array has less protection after one failure, but can remain protected against another failure.
• A RAID 10 rebuild generally copies the surviving mirror member to the replacement drive.
• Heavy Plex activity, surveillance writes, backups, and file transfers can extend the process.
• The rebuild itself places sustained read and write demand on the remaining drives.

A larger drive does not automatically make a layout unsafe, but it increases the amount of data that must be reconstructed. This is one reason to compare RAID 5 with RAID 6, RAIDZ1 with RAIDZ2, and parity layouts with RAID 10 rather than choosing only by capacity.

Hot spare versus cold spare

A hot spare is an installed drive reserved for automatic use when a member fails. It can shorten the time before rebuilding begins, but it does not increase the array's normal usable capacity or change the RAID level's failure tolerance before it is activated.

A cold spare is a compatible replacement kept available but not installed in the array. It avoids using a drive for standby service, but someone must identify the failure and replace the drive manually.

Consider a hot spare when:

• The NAS is remote or difficult to access.
• Downtime matters.
• The platform supports automatic spare activation as expected.
• You can match the spare's capacity and relevant drive requirements.

Consider a cold spare when:

• You can respond quickly to alerts.
• Budget or drive-bay count is limited.
• You want the spare stored separately from the NAS.
• You are comfortable validating compatibility before installation.

A spare is not a backup. It protects against a future drive failure in the array; it does not preserve deleted, corrupted, encrypted, or overwritten files.

4. Check expansion flexibility before buying the first drives

Expansion is often more complicated than replacing one failed drive with a larger one.

Ask these questions before committing to a layout:

• Can the NAS expand the array by replacing drives one at a time?
• Must every drive be replaced before the additional capacity becomes available?
• Can the NAS add a new RAID group or vdev?
• Does expansion require a particular RAID level or filesystem?
• Can the NAS expand only by adding a complete group of drives?
• Will a new group have different performance or failure characteristics?
• Is expansion supported by the NAS manufacturer for the exact model and storage mode?

RAID 10 commonly expands by adding mirror pairs, where supported. Traditional RAID 5 and RAID 6 expansion behavior varies by controller, NAS operating system, and filesystem. ZFS pool expansion depends on the ZFS implementation and vdev design; adding drives later is not equivalent to changing every existing vdev.

If you expect substantial growth, compare the cost of:

• Buying a larger initial array
• Adding another complete group later
• Replacing every drive with a larger model
• Moving data to a new pool or NAS
• Using a separate expansion enclosure, if supported

A storage layout that is slightly cheaper today can be more expensive if its only practical upgrade path is a full migration.

5. Match the RAID choice to the workload

Capacity and failure tolerance are only part of the decision.

File storage and documents

For ordinary file storage, choose based on protection, capacity, and backup requirements. Small arrays may make RAID 1 attractive because it is simple to understand and provides one-drive tolerance. Larger arrays may justify dual parity when rebuilding and downtime risk matter more.

Plex and media libraries

Plex usually benefits more from adequate network connectivity, CPU capability for transcoding, and enough storage than from a particular RAID level. RAID can keep the media share online after a drive failure, but it does not protect against accidental deletion or a damaged media library database.

Surveillance recording

Surveillance workloads write continuously and can fill an array quickly. Calculate retention from camera count, resolution, frame rate, codec, motion settings, and retention days rather than assuming a generic per-camera figure. RAID helps maintain availability after a drive failure, but a separate backup or export strategy is still needed for footage that must be retained.

SSD cache

SSD cache is not a substitute for more hard-drive capacity or a sound RAID layout. It can add complexity, consume drive bays, and have workload-specific benefits. Decide on the primary HDD array first. Add cache only when the NAS software, workload, endurance requirements, and failure behavior are understood.

6. Worked example: planning a 20 TB target

Suppose the goal is to store 20 TB of data and retain approximately 20% free space.

First calculate the nominal capacity target:

20 TB / 0.80 = 25 TB

Assume 8 TB drives are available. The following options illustrate the trade-offs:

Layout	Drive count	Nominal RAID capacity	Capacity after a 20% planning reserve	Failure tolerance
RAID 5	4 × 8 TB	24 TB	19.2 TB	One drive
RAID 5	5 × 8 TB	32 TB	25.6 TB	One drive
RAID 6	5 × 8 TB	24 TB	19.2 TB	Any two drives
RAID 6	6 × 8 TB	32 TB	25.6 TB	Any two drives
RAID 10	6 × 8 TB	24 TB	19.2 TB	One drive per mirror pair, depending on failures
RAID 10	8 × 8 TB	32 TB	25.6 TB	One drive per mirror pair, depending on failures

In this example, the 24 TB layouts do not meet the 25 TB nominal planning target once the 20% reserve is applied. The 32 TB layouts do, subject to filesystem overhead and the NAS's actual reported capacity.

The choice then becomes practical:

• Five-drive RAID 5 uses fewer drives but has only single-drive tolerance.
• Six-drive RAID 6 uses an additional drive for dual-parity protection.
• Eight-drive RAID 10 uses more bays and raw capacity but has mirror-based rebuild behavior.
• A hot spare would require another compatible drive and bay or an external replacement plan; it does not increase the usable capacity shown above.

If the NAS has limited bays, the expansion path may matter more than the initial difference in drive count. Confirm the platform's supported expansion method before buying.

7. RAID is not backup

RAID protects against some drive failures. It does not protect against:

• Accidental deletion
• Ransomware or compromised accounts
• File corruption copied across the array
• Theft, fire, flood, or power-related damage
• A failed NAS enclosure or controller
• Incorrect administrative changes
• A user deleting the only copy of a file

Use a separate backup system with a recovery process you have tested. A backup should be independent enough that an array failure or NAS compromise does not destroy both the primary data and its backup.

A useful planning rule is:

• RAID provides availability.
• Backup provides recoverability.
• A spare reduces the time needed to restore redundancy.
• Snapshots can help with recent changes, but they are not automatically an independent backup.

Before buying drives, verify that your backup target can hold the data that matters and that the NAS can complete backups within your required schedule.

RAID decision table

If your priority is...	Usually start by considering...	Main trade-off
Maximum capacity with no failure tolerance	RAID 0	One drive failure can lose the array
Simple protection in a small NAS	RAID 1	Usable capacity is about one drive
More usable capacity with one-drive tolerance	RAID 5 or RAIDZ1	Reduced protection during and after a failure
Two-drive tolerance	RAID 6 or RAIDZ2	More capacity consumed by parity
Mirror-based rebuilds and predictable redundancy	RAID 10	About half of raw capacity is usable; failure pattern matters
Remote operation with faster automatic response	Chosen RAID level plus a hot spare	Extra drive cost; no additional backup protection
Fast manual replacement at lower idle cost	Chosen RAID level plus a cold spare	Someone must respond to the failure
Frequent future growth	A layout with a documented expansion path	Expansion rules vary by NAS, RAID implementation, and filesystem
Critical data	Redundant RAID plus tested independent backups	RAID alone does not provide recovery from deletion or disaster

Before purchasing drives, write down the required usable capacity, acceptable number of simultaneous failures, planned free-space reserve, NAS bay count, expansion method, spare strategy, and backup destination. Then compare compatible NAS hardware through NAS and storage servers or Browse NAS.