RAID Capacity and Sizing Calculator

Current layout is not buildable from these inputs.

{{ issue }}

{{ warning }}

RAID layout:

Supported layouts cover RAID 0, 1, 5, 6, 10, 50, and 60 for capacity planning.

Installed drives: {{ normalizedInput.installedDrives }} drives

Use the planned drive count in the shelf or array.

drives

Drive size:

Choose TB/GB for decimal vendor sizes or TiB/GiB for binary operating-system values.

Dedicated hot spares:

Use 0 when every installed drive is active in the RAID set.

spares

Parity groups:

Each group must have enough active drives for the selected nested parity layout.

groups

Target usable capacity:

Set the capacity the finished array should safely provide to workloads.

Reserve headroom: {{ formatPercent(normalizedInput.reservePct, 0) }}

Common planning targets keep 10-20% unallocated; 0% means no safety headroom.

Metadata/system overhead:

Leave at 0 for generic RAID math; set a vendor or filesystem reserve when known.

Chassis slot limit:

Use 0 when no chassis or expansion limit should be checked.

slots

Cost per drive:

Leave at 0 to omit cost guidance from the ledger and JSON.

Display precision:

Capacity math uses bytes internally; this only changes displayed rounding.

Metric	Value	Read	Copy
{{ row.metric }}	{{ row.value }}	{{ row.read }}

Layout	Planned usable	Efficiency	Fault tolerance	Target read	Copy
{{ row.layout }}	{{ row.plannedUsable }}	{{ row.efficiency }}	{{ row.faultTolerance }}	{{ row.targetRead }}

Path	Requirement	Result	Status	Copy
{{ row.path }}	{{ row.requirement }}	{{ row.result }}	{{ row.status }}

{{ jsonOutput }}

Embed:

Customize

Include current inputs

Size

Advanced

Width

Height

Aspect ratio

Max height

Collapsible embed

Allow fullscreen

Referrer policy

Sandbox tokens

A drive shelf is bought in raw terabytes, but applications live on the smaller amount that remains after protection, standby drives, system overhead, and operating margin. RAID capacity sizing is the practice of turning the physical drive plan into a workload-ready number before the purchase order, migration window, or expansion slot count becomes fixed.

RAID, short for redundant array of independent disks, combines drives into a logical storage unit by striping, mirroring, parity, or a nested mix of those ideas. RAID 0 keeps the most capacity because it stores no redundancy. RAID 1 and RAID 10 trade capacity for mirrored copies. RAID 5, RAID 6, RAID 50, and RAID 60 reserve one or more drive equivalents for parity so the array can reconstruct data after specific drive failures.

The planning question is not just how many bytes fit on the shelf. A file server with snapshots, a virtualization datastore, and an archive appliance can all use the same 12 TB disks and end up with different safe targets because the RAID level, hot-spare policy, metadata reserve, and desired unused-space cushion change the answer. Decimal drive labels and binary operating-system displays add another source of confusion: 12 TB is not 12 TiB.

Common RAID capacity planning terms
Planning term	Meaning	Why it changes usable space
Raw capacity	Installed drives multiplied by the advertised drive size.	It is the starting quantity, not the workload-ready quantity.
Active drives	Drives left after dedicated hot spares are removed.	Only active drives can contribute data, parity, or mirror capacity.
Protection capacity	Capacity spent on parity blocks or mirrored copies.	It improves drive-failure tolerance but lowers efficiency.
Reserve headroom	Usable space intentionally left unused.	It gives room for growth, snapshots, and operational policy.

RAID capacity flow from installed drives through spares, active drives, parity or mirrors, overhead, and reserve to planned usable capacity

The most expensive mistake is counting every drive label twice: once in the raw total and again in the promised usable total. Hot spares are not active members, parity and mirrors consume drive equivalents, and reserve headroom should stay available if snapshots, thin provisioning, or growth are part of the operating plan.

A target-fit capacity result still leaves design work to do. RAID improves local availability for some drive failures, but it does not protect against deletion, corruption, ransomware, controller faults, site loss, or untested restores. Treat the capacity number as one input to a storage design that also needs backup, monitoring, rebuild policy, and vendor support checks.

How to Use This Tool:

Start with the shelf geometry, then compare the planned usable result with the target and the sizing path.

Choose RAID layout. Supported choices are RAID 0, RAID 1, RAID 5, RAID 6, RAID 10, RAID 50, and RAID 60.
Enter Installed drives, Drive size, and the drive size unit. Use TB or GB for decimal vendor labels, and TiB or GiB when the source number is already binary.
Set Dedicated hot spares. These drives are removed before capacity math because they are standby drives, not normal data or parity members.
For RAID 50 or RAID 60, set Parity groups. The group count must leave enough active drives per group, or the tool reports that the layout is not buildable.
Enter Target usable capacity and Reserve headroom. Open Advanced when you need to include metadata/system overhead, a chassis slot limit, drive cost, or display precision.
Check the summary badges for target met, target short, drive-failure tolerance, active drive count, and installed efficiency. If a warning or fatal issue appears, fix that input before treating the plan as usable.
Use Sizing Ledger for the current build, RAID Alternatives for same-drive comparisons, Capacity Mix Chart for where raw capacity goes, and Drive Count Ladder or Sizing Path when the target is short.

Interpreting Results:

Planned usable capacity is the main capacity answer. It is calculated after the selected RAID layout, dedicated hot spares, metadata/system overhead, and reserve headroom are applied. Compare that value with the target badge and the Planned usable capacity row in the ledger.

Installed efficiency shows how much of the installed raw capacity remains planned usable. Low efficiency is not automatically bad. RAID 10, RAID 6, hot spares, and reserve headroom can be deliberate choices when reliability, write behavior, or operating margin matters.

Minimum drive count and Required drive size answer different sizing questions in Sizing Path. Minimum drive count keeps the selected drive size and looks for the first drive count that meets the target. Required drive size keeps the current drive count and estimates how large each drive would need to be.

A target met badge does not certify the storage design. Check Drive-failure tolerance, idle or uneven capacity warnings, chassis slot pressure, backup coverage, controller support, and rebuild policy before procurement.

Technical Details:

RAID capacity sizing is driven by drive equivalents. Each layout converts active drives into data-drive equivalents and protection-drive equivalents. Data-drive equivalents produce usable capacity before overhead; protection-drive equivalents represent parity or mirrored copies that improve fault tolerance but are not available as workload capacity.

The arithmetic assumes equal effective drive sizes. In mixed-drive arrays, many RAID systems size the set as if every member were the smallest usable drive, so a planning estimate should use that common effective size rather than the largest label on the shelf.

Formula Core:

The core equation starts with installed raw capacity, removes inactive and protection capacity through the selected data-drive rule, then applies overhead and reserve percentages.

\begin{array}{lcl} C_{raw} & = & N \times S \\ N_{active} & = & N - H \\ C_{pre} & = & D \times S \\ C_{planned} & = & C_{pre} \times (1 - \frac{o}{100}) \times (1 - \frac{r}{100}) \\ Installed efficiency & = & \frac{C_{planned}}{C_{raw}} \times 100 % \end{array}

Here N is installed drive count, H is dedicated hot spares, S is drive size in bytes, D is data-drive equivalents from the RAID rule, o is metadata/system overhead percent, and r is reserve headroom percent. Display rounding is applied after byte-level arithmetic.

RAID layout data-drive rules
Layout	Data-drive equivalent rule	Drive-failure tolerance shown	Capacity note
RAID 0	`active drives`	`0 drives`	No parity or mirror capacity is reserved.
RAID 1	`1`	`active drives - 1`	One drive equivalent is usable; the rest hold identical copies.
RAID 5	`active drives - 1`	`1 drive`	One drive equivalent is spent on distributed parity.
RAID 6	`active drives - 2`	`2 drives`	Two drive equivalents are spent on dual distributed parity.
RAID 10	`engaged even drives / 2`	`1 per mirror pair`	An odd active drive cannot form a mirrored pair and is treated as idle.
RAID 50	`groups x (group width - 1)`	`1 per group`	Each parity group behaves like a RAID 5 set, then groups are striped together.
RAID 60	`groups x (group width - 2)`	`2 per group`	Each parity group behaves like a RAID 6 set, then groups are striped together.

Grouped parity layouts use whole groups. If 25 active drives are split into 4 RAID 60 groups, each group receives 6 drives and one active drive is left outside the modeled layout. That leftover capacity appears as idle or uneven capacity because it cannot be evenly assigned to the selected group count.

Worked substitution: 8 installed 12 TB drives in RAID 6 with no hot spares have 6 data-drive equivalents. Capacity before overhead and reserve is 72 TB. With 0% metadata/system overhead and 10% reserve headroom, planned usable capacity is 64.8 TB. Against a 60 TB target, the result is target met with 4.8 TB of margin before filesystem-specific effects.

Input bounds and validation behavior
Input	Accepted range	Validation effect
Installed drives	`2` to `120`	Must also satisfy the selected RAID layout minimum after hot spares.
Dedicated hot spares	`0` to one less than installed drives	Removed before active capacity, parity, and mirror rules are applied.
Drive size	`0.01` to `10000` in the selected unit	Converted to bytes using decimal or binary unit multipliers.
Reserve headroom	`0%` to `50%`	Applied after metadata/system overhead.
Metadata/system overhead	`0%` to `25%`	Applied before reserve headroom.
Parity groups	`2` to `40`	RAID 50 needs at least 3 active drives per group; RAID 60 needs at least 4.

Accuracy Notes:

Capacity math is a planning estimate. It helps compare RAID layouts and drive counts, but the final usable number can change once a real controller, filesystem, volume manager, snapshot policy, or storage platform applies its own rules.

Mixed-size arrays often behave as if every member were the smallest effective drive.
Decimal drive labels and binary operating-system displays are different quantities. A 12 TB drive is smaller than 12 TiB.
Compression, deduplication, thin provisioning, snapshots, and vendor reserve can make observed capacity differ from generic RAID math.
RAID is local redundancy, not backup. Keep separate recovery copies for deletion, corruption, ransomware, and site failure.
Large drives and wide parity groups can increase rebuild time and degraded-operation exposure, even when the capacity result meets the target.

Worked Examples:

Dual-parity file storage. Eight 12 TB drives in RAID 6, no hot spares, a 60 TB target, and 10% reserve produce 64.8 TB of Planned usable capacity. The summary should read target met, while Drive-failure tolerance reads 2 drives.

Mirrored database shelf. Ten active 7.68 TB SSDs in RAID 10 produce five data-drive equivalents before reserve. With 15% reserve and no metadata/system overhead, Planned usable capacity is about 32.64 TB; a 30 TB target is met, but Installed efficiency is far lower than parity RAID because half the engaged drives are mirrors.

Grouped parity with an uneven count. Twenty-five active 4 TB drives in RAID 60 across four parity groups form four six-drive groups and leave one active drive outside the grouped layout. The ledger includes Idle or uneven active capacity, so resolve the group count or drive count before treating the shelf as fully used.

Not enough drives for the selected groups. Ten active drives with RAID 60 and four parity groups cannot be built because four groups need at least sixteen active drives. Reduce Parity groups, add drives, or choose another RAID layout until the fatal issue disappears.

Advanced Tips:

Use the smallest effective drive size when a planned array has mixed models or mixed usable capacities; otherwise the ledger can overstate usable space.
Keep Reserve headroom separate from Metadata/system overhead. Reserve is capacity you choose not to allocate, while overhead models space consumed before that reserve is held back.
For RAID 50 and RAID 60, adjust Parity groups until the ledger no longer reports idle or uneven active capacity unless the unused drive is intentional.
Set Chassis slot limit before relying on Minimum drive count; a fit that needs more bays than the enclosure has is a procurement problem, not a math success.
Use RAID Alternatives to compare efficiency and drive-failure tolerance together. A layout with more planned usable capacity may still be a poor fit for the rebuild exposure or workload risk.

FAQ:

Why is planned usable capacity lower than raw capacity?

Raw capacity counts every installed drive at its label size. Planned usable capacity removes dedicated hot spares, parity or mirror allocation, metadata/system overhead, and reserve headroom.

Why do TB and TiB produce different results?

TB and GB use decimal powers of 1000. TiB and GiB use binary powers of 1024. The same byte count is displayed as a smaller number in TiB than in TB.

What should I do when the layout is not buildable?

Read the fatal issue first. The usual fixes are adding active drives, lowering dedicated hot spares, reducing RAID 50 or RAID 60 parity groups, or choosing a RAID layout with a lower drive minimum.

Does RAID 10 always tolerate more than one failed drive?

RAID 10 can survive multiple failures only when they land in different mirrored pairs. The tool reports 1 per mirror pair because two failed drives in the same pair can still break the array.

Does meeting the target mean I can skip backups?

No. A target-fit result only says the modeled RAID plan has enough planned usable capacity. Backups, replication, snapshots, monitoring, and recovery testing are separate requirements.

Glossary:

Raw capacity: Total installed drive capacity before hot spares, RAID protection, overhead, and reserve are removed.
Active drive: A drive that remains in the RAID layout after dedicated hot spares are excluded.
Data-drive equivalent: The number of drive-size units that contribute to usable capacity before overhead and reserve.
Parity group: A RAID 5 or RAID 6 subgroup used inside RAID 50 or RAID 60.
Hot spare: A standby drive reserved for rebuild workflow rather than normal data, parity, or mirror capacity.
Reserve headroom: Usable capacity intentionally left unallocated for growth, snapshots, policy, or operating margin.

References:

RAID level summary, IBM Power documentation, 12 May 2022.
Managing RAID, Red Hat Enterprise Linux documentation.
Supported Levels for Intel RAID Controllers, Intel, last reviewed 03 Sep 2024.
What is Erasure Coding?, SNIA.