RAID Calculator – Complete Guide to Usable Capacity, Fault Tolerance and RAID Levels

Before you order twelve 16 TB drives, you want to know how much of that 192 TB you will actually be able to use, how many disks can die before the data is gone, and what the choice of RAID level does to write performance. Vendor sizing tools usually cover only their own hardware; mental arithmetic breaks down with nested levels, unequal disks and decimal-versus-binary units.

The RAID Calculator calculates raw, usable and available capacity, storage efficiency, theoretical performance multipliers and fault tolerance for RAID 0, 1, 1E, 5, 5E, 5EE, 6, 10, 50 and 60. It draws a logical disk layout and lets you click disks to simulate failures. Everything runs in your browser, from one exact model.

What is the RAID Calculator?

In simple terms

You choose a RAID level, the number of disks and their size; the calculator tells you how much space you get, how much is spent on parity or mirror copies, how many disk failures the array survives, and roughly how it performs compared with a single disk.

Technical background

RAID (Redundant Array of Independent Disks) combines disks into one logical volume. The classic techniques, described in the 1988 Berkeley paper by Patterson, Gibson and Katz and since implemented by every controller and software stack, are:

• Striping (RAID 0): data is split across disks. Capacity and throughput add up; there is no redundancy.
• Mirroring (RAID 1): every disk holds a full copy.
• Parity (RAID 5, RAID 6): one or two disks' worth of parity — XOR-based for single parity, a second independent syndrome for RAID 6 — is distributed across all disks, so the data of a failed disk can be recomputed.
• Nesting (RAID 10, 50, 60): mirror sets or parity groups striped together.

Two rules drive capacity. First, every disk in an array — or in each group of a nested array — is used only up to the size of its smallest member. Second, redundancy costs capacity: mirrors store copies, parity levels store one or two disks' worth of parity, and spare-capacity levels hold room for a rebuild.

RAID is not a backup. It protects against some disk failures only. Deletions, corruption, ransomware and controller faults are replicated to every disk instantly.

Why use the RAID Calculator?

• Size purchases correctly: see usable capacity in decimal TB (as drives are labelled) and binary TiB (as many operating systems show it).
• Compare levels for the same disks: RAID 6 versus RAID 10 versus RAID 60 in capacity, write penalty and tolerance.
• Plan nested arrays with valid subgroup splits for RAID 50 and 60.
• See the cost of unequal disks — the calculator lists exactly how much capacity sits unused.
• Test failure scenarios by clicking disks instead of reasoning about which pair or group a disk belongs to.
• Document a design with CSV, JSON, text or a printable report including the layout diagram.

Who should use it?

System and storage administrators building servers or NAS arrays, MSPs quoting storage, homelab users choosing a NAS layout, and anyone reviewing a storage design. Pair it with the Data Storage Converter when you need to translate the result into the units a particular tool reports.

How the RAID Calculator works

The calculator runs entirely in your browser. All capacities are exact rational numbers of bytes, so the segments always add up to the raw capacity.

Capacity model

raw        = sum of all disk sizes
unused     = sum of (disk size − smallest disk of its group)
usable     = raw − unused − parity − mirror copies − spare
reserved   = usable × reservation %
available  = usable − reserved
efficiency = usable ÷ raw (rounded to 2 decimals)

The levels are modelled as follows (N = disks, S = smallest disk, G = groups, g = disks per group, w = copies per mirror set, s = spare disks' worth):

With unequal disks, the smallest-disk rule is applied per mirror set (RAID 10) or per group (RAID 50/60), so the totals are sums over groups.

Vendor-specific levels

RAID 1E, 5E and 5EE are not standardized, and the calculator says which layout it models:

• RAID 1E — the "adjacent" layout: each stripe block lives on one disk and is mirrored on the next disk, wrapping from the last disk to the first. Any disk count from 3 works. Losing two neighboring disks (including last and first) loses data.
• RAID 5E — RAID 5 plus spare capacity equal to one or two disks, kept as an unused region at the end of every disk.
• RAID 5EE — RAID 5 with the spare blocks distributed through every stripe, like parity, so all disks serve reads and rebuilds use all spindles.

For 5E and 5EE the simultaneous tolerance is that of RAID 5. After a rebuild into the spare capacity the array can absorb another failure (two more, one at a time, with two spares' worth). At least two disks' worth of data must remain, so two spares need at least 5 disks.

Validation

The calculator refuses impossible configurations with an explanation: more than 64 disks, sizes of zero, fewer disks than the level's minimum, a reservation outside 0–50%, RAID 10 disk counts that are not a multiple of the mirror width or form fewer than two mirror sets, RAID 50/60 groups smaller than 3 (RAID 5) or 4 (RAID 6) disks, unequal groups, or a single group (which would just be RAID 5 or 6).

Groups and layout

Groups are formed from consecutive disks: RAID 10 pairs are disks 1+2, 3+4 and so on; RAID 50/60 groups are disks 1–g, g+1–2g, etc. The layout diagram shows each disk's share of data, parity, mirror copy, spare and unused space as a logical allocation. Real controllers rotate parity and mirror blocks across disks; the diagram shows amounts, not physical block positions.

Performance model

The multipliers are theoretical upper bounds for large sequential transfers relative to one member disk; no MB/s or IOPS are claimed. The write penalty is the number of disk I/Os per small random write: 4 for single parity (2 reads + 2 writes), 6 for double parity (3 reads + 3 writes), w for a w-way mirror.

Warnings

The calculator always adds "RAID is not a backup". It also warns about unequal disk sizes (with the unused amount), about RAID 0's lack of redundancy, about RAID 5 and RAID 50 when any disk is 8 TB (decimal) or larger, and about the vendor-specific nature of 1E, 5E and 5EE.

How to use the NetTools.tech RAID Calculator

1. Open the RAID Calculator.
2. Pick the RAID level. The hint below shows its name, minimum disks and the modelled layout.
3. Enter the Number of disks (1–64), the Disk size and the Unit: decimal GB, TB, PB (as on drive labels) or binary GiB, TiB, PiB.
4. Tick Disks have different sizes to enter each disk individually (Disk 1, Disk 2, …) in the selected unit.
5. Set the level-specific option: Subgroups for RAID 50/60 (only valid splits into at least two equal groups are offered, such as "2 groups × 6 disks"), Mirror layout for RAID 10 (Pairs or Triple mirrors) or Spare capacity for RAID 5E/5EE (1 or 2 disks' worth). Options for other levels are shown disabled.
6. Optional: enter a System reservation (% of RAID capacity, 0–50) for filesystem overhead, metadata, snapshots or system partitions.
7. Choose Show capacities in Decimal (GB, TB) or Binary (GiB, TiB); every value also shows the other system as a note.
8. Or click an example: 4 × 8 TB RAID 5, 6 × 12 TB RAID 6, 8 × 4 TB RAID 10, 12 × 16 TB RAID 60, 2 × 2 TB mirror.
9. In Disk layout, click disks to mark them as failed; the simulator states whether the array survives. Clear failures resets it.
10. Export with CSV, JSON, Copy all or Print. The print report includes the capacity bar, the disk layout and any failure simulation.

Understanding the results

Example: four 8 TB disks in RAID 5, then RAID 10 failures

RAID 5, 4 × 8 TB, 10% reservation, decimal display:

Raw capacity           32 TB        (29.1 TiB)
RAID-usable capacity   24 TB        (21.83 TiB)
Reserved for system    2.4 TB       (10% of usable)
Available capacity     21.6 TB      (19.65 TiB)
Parity                 8 TB
Storage efficiency     75 %
Capacity bar           reserved 7.5% · available 67.5% · protection 25% · unused 0%
Each disk              6 TB data · 2 TB parity
Sequential read ×3 · Sequential write ×3 · Random write penalty 4
Guaranteed 1 disk · Best case 1 disk

Because the disks are 8 TB, the calculator warns that RAID 5 rebuilds take long and that RAID 6 is often preferred. Mark disk 1 as failed: "Array survives — 1 simultaneous failure (disk 1). One failure: data is rebuilt from parity." Add disk 2: "Array lost … 2 simultaneous failures exceed single parity."

RAID 10, 8 × 4 TB, pairs: usable 16 TB, efficiency 50%, read ×8, write ×4, penalty 2, guaranteed 1, best case 4. Fail disks 1, 3, 5 and 7: the array survives, because every mirror set still has a copy. Fail disks 1 and 2: "Mirror set 1 lost 2 disks (tolerates 1); the whole array fails."

Unequal disks, RAID 5 with 4, 8, 8 and 12 TB: usable is only 12 TB (3 × 4 TB), 16 TB are unused, and efficiency drops to 37.5%.

RAID 1E with 5 × 4 TB: 10 TB usable, best case 2. Failing disks 1 and 3 survives; disks 1 and 5 are neighbors (the layout wraps around) and the array is lost.

Common use cases

• Choosing a RAID level for a new server or NAS from the same set of disks.
• Checking the usable capacity a quote promises, in TB and TiB.
• Planning RAID 50/60 group sizes for large arrays.
• Evaluating whether replacing one disk with a larger model gains anything (it does not, until all members of the group are larger).
• Explaining fault tolerance to stakeholders with the failure simulator and a printed report.

Troubleshooting

"This configuration is not possible"

Read the listed reason. Typical fixes: use an even disk count for RAID 10 pairs (a multiple of 3 for triple mirrors), at least 6 disks for RAID 50 or 8 for RAID 60, a disk count that splits into equal groups, or at least 5 disks for 5E/5EE with two spares.

"No valid split for N disks" in Subgroups

RAID 50/60 needs at least two equal groups of at least 3 or 4 disks. Prime disk counts such as 7 or 11 have no valid split; change the disk count.

The operating system reports less space than the calculator

Switch Show capacities in to Binary. If the difference remains, add a system reservation for filesystem metadata, reserved blocks, snapshots and system partitions.

The large-disk warning appears although my usable disk size is small

The warning checks whether any disk in a RAID 5 or RAID 50 configuration is 8 TB or larger, including larger disks whose extra capacity is unused.

Common mistakes

• Treating RAID as a backup. Keep independent, ideally offline or off-site, backups.
• Planning with the best-case tolerance. Failures do not choose convenient disks; plan with the guaranteed number.
• Ignoring rebuild risk. During a RAID 5 rebuild there is no redundancy left, and rebuilds of large HDDs take a long time.
• Mixing disk sizes and expecting the sum. Only the smallest disk per group counts.
• Reading performance multipliers as throughput. They are theoretical ratios; controller, cache, HDD versus SSD and workload decide real numbers.
• Assuming 1E/5E/5EE behave identically on every controller. Check your vendor's documentation.

Limitations

• The model is classic RAID. ZFS RAID-Z, Btrfs, hybrid NAS schemes, erasure coding and dynamic disk pools allocate capacity differently.
• Performance figures are theoretical and relative to one disk; there are no MB/s, IOPS or rebuild-time estimates, and no reliability (MTBF, URE) calculation.
• The failure simulator tests simultaneous failures only; it does not simulate rebuilds into spare capacity for 5E/5EE, unrecoverable read errors or controller failure.
• Vendor-specific levels follow the documented model, which may differ from your controller.
• Available capacity is an estimate based on your reservation percentage.

Security & privacy

The calculator runs locally in your browser; disk counts, sizes and results are not sent to a server. The page may send an anonymous usage counter (tool name and view or run, no inputs) unless your browser uses Global Privacy Control or Do Not Track. Exports are generated in the browser and saved only where you put them.

Related tools

• Data Storage Converter — TB versus TiB and every other storage unit. Guide: Data Storage Converter guide.
• Bandwidth Calculator — how long it takes to back up or replicate the array. Guide: Bandwidth Calculator guide.
• Unit Converter — data, rates and time.
• Bits & Bytes Explainer — decimal versus binary prefixes. Guide: Bits and bytes guide.
• Category overview: Conversion & calculation tools.

Summary

The RAID Calculator applies one exact model to ten RAID levels: smallest disk per group, redundancy subtracted, optional reservation, efficiency as usable ÷ raw. It shows guaranteed and best-case fault tolerance, theoretical performance ratios and a clickable failure simulator, and it labels vendor-specific levels as such. Use it to compare designs — and keep backups, because RAID is not one.