S
SoSimple.tools
Tech & Hardware

RAID Calculator for NAS: Usable Capacity & Rebuild Risk

Find out how much space your NAS array really gives you after parity, how much your operating system will display, and how risky a rebuild is with large hard drives

Runs in your browser: what you enter stays on your device

Array Configuration

4 drives

Total drives in the NAS, including any hot spares

Array Results
Usable Capacity
24 TB
21.83 TiB · as shown by your OS
Raw Capacity
32 TB
Space Efficiency
75%
Drive Failures Survived
1 drive
Rebuild Time
22.2 h
Chance of a read error during rebuild 85.3%

The rebuild is the weak point: with this drive size and datasheet error rate, hitting a read error mid-rebuild is plausible. Consider dual parity (RAID 6, RAIDZ2, SHR-2) and keep a separate backup.

Share this tool

How to use this tool

  1. Choose the RAID level (including SHR and RAIDZ) and the drive capacity and number of drives, or tick Mixed drive sizes and list them.
  2. Optionally add hot spares, pick a rebuild speed and the drive's unrecoverable read error rate (URE).
  3. Read the usable capacity, the figure your OS will show in TiB, the efficiency and how many drives can fail.
  4. Check the rebuild time and the chance of a read error during a rebuild.

Worked example

Four 8 TB drives in RAID 5 give (4 − 1) × 8 = 24 TB usable out of 32 TB raw (75%), shown as 21.83 TiB by an OS that counts in binary units. Rebuilding one 8 TB drive at about 100 MB/s takes 22.2 hours, during which the array reads 3 × 8 TB = 1.92 × 10¹⁴ bits. With one error per 10¹⁴ bits that is 1 − e^(−1.92) = 85.3% chance of at least one read error, but only about 17.5% at one per 10¹⁵ bits. With mixed drives of 8, 8, 4 and 4 TB, RAID 5 truncates every drive to 4 TB and yields 12 TB, while SHR-1 yields 16 TB. These are planning estimates; datasheet error rates are upper limits and RAID is not a backup.

Methodology

Capacity formulas follow the parity scheme introduced in the 1988 Berkeley paper by Patterson, Gibson and Katz, A Case for Redundant Arrays of Inexpensive Disks, which defined RAID levels 1 to 5: RAID 5 spreads one drive's worth of parity across the array. RAID 6 and RAIDZ2 are later dual-parity designs that store two drives' worth. Everything is computed locally in your browser.

Capacity rules used: RAID 0 = n×s · RAID 1 = s · RAID 5 and RAIDZ1 = (n−1)×s · RAID 6 and RAIDZ2 = (n−2)×s · RAID 10 = n/2×s, where s is the smallest drive. With mixed sizes, classic levels truncate every drive to s. SHR slices drives into equal-size layers and builds redundancy per layer, so SHR-1 yields total capacity minus the largest drive. Operating systems report binary units: 1 TiB = 2^40 bytes (IEC 80000-13), while drive makers count 1 TB = 10^12 bytes.
Rebuild-risk model: Chance of at least one unrecoverable read error during a rebuild = 1 − e^(−B/U), where B is the number of bits read from the surviving drives and U is the datasheet error interval. Rebuild time = failed-drive size ÷ sustained rebuild rate. Datasheet error rates are specified as upper limits, so real-world risk is often lower. ZFS resilvering copies only allocated blocks, so a half-full RAIDZ pool reads less than this full-drive model assumes. Treat every figure as a planning estimate, not a guarantee.

Four 12 TB drives, four layouts, and a mixed-drive case

Four 12 TB drives, rebuild at about 100 MB/s (33.3 h per failed drive), error rate 1 per 10^14 / 10^15 bits. Usable space is in TB / TiB with efficiency in brackets.

LevelUsable TB / TiBFailures survivedRead error during rebuild (10^14 / 10^15)
RAID 5 / RAIDZ136 / 32.74 (75%)1 drive94.4% / 25.0%
RAID 6 / RAIDZ224 / 21.83 (50%)2 drivesCovered by second parity
RAID 1024 / 21.83 (50%)1 guaranteed (up to 2)61.7% / 9.2%
RAID 048 / 43.66 (100%)NoneNot applicable

Case: Inês has five old drives of 10, 10, 6, 6 and 4 TB (36 TB raw). RAID 5 treats all of them as 4 TB: (5 − 1) × 4 = 16 TB usable, with 16 TB left unused. SHR-1 slices the drives into layers and delivers 26 TB (23.65 TiB, 72.2%). The price is a longer rebuild: the largest drive is 10 TB, so about 27.8 h at 100 MB/s, with an 87.5% chance of a read error at 1 per 10^14 bits but 18.8% at 1 per 10^15. Her choice is SHR-2 (16 TB usable, covered by second parity) if the data matters, plus a separate backup. These are planning estimates; RAID is not a backup.

Build Your NAS: Drives, Enclosure & Power Protection

Three building blocks for a reliable home or small-office storage array

NAS Hard Drive Array Storage
Seagate IronWolf 8 TB NAS Hard Drive (ST8000VNZ04)

Seagate IronWolf 8 TB NAS Hard Drive (ST8000VNZ04)

Built for 24/7 NAS use: all-CMR recording, RV sensors for multi-bay enclosures, a 180 TB/year workload rating, and a 3-year warranty plus 3-year Rescue data recovery.

2-Bay NAS Home Server
UGREEN NAS DXP2800 2-Bay (Diskless)

UGREEN NAS DXP2800 2-Bay (Diskless)

Intel N100, 8 GB DDR5, 2.5GbE, two M.2 NVMe slots and 4K HDMI, suited to Docker, VMs, Plex and fast file transfer. Sold without drives.

Battery Backup Power Protection
APC BE600M1 UPS 600VA Battery Backup

APC BE600M1 UPS 600VA Battery Backup

600VA/330W backup with 490 J surge protection and USB charging, rated for about 23 minutes at 100 W so your NAS can shut down cleanly during an outage.

As an Amazon Associate we earn from qualifying purchases at no additional cost to you.

Related tools

Frequently asked questions

Why does my NAS show about 21.8 TB when I built 4 × 8 TB in RAID 5? ▼

Two separate effects stack. RAID 5 gives one drive to parity, leaving 24 TB in decimal units. Your operating system then reports binary units: 24 × 10^12 bytes is about 21.83 TiB, often still labeled “TB”. Filesystem metadata takes a further small slice.

Is RAID 5 still safe with 8 TB or larger drives? ▼

It depends on your drive's datasheet error rate and your backups. Rebuilding a 4 × 8 TB RAID 5 means reading 24 TB from the survivors; at 1 error per 10^14 bits the model puts the chance of hitting a read error near 85%, a worst-case figure that real drives usually beat. Dual parity (RAID 6, RAIDZ2, SHR-2) removes that single point of failure, and no RAID level replaces a backup.

What happens if I mix a 4 TB drive and an 8 TB drive? ▼

Classic RAID 1/5/6/10 and RAIDZ treat both as 4 TB and strand the extra 4 TB. SHR slices the larger drive so most of that space stays usable. Tick “Mixed drive sizes”, enter your own sizes and compare SHR-1 against RAID 5.

Does this calculator subtract ZFS or filesystem overhead? ▼

No, it shows the theoretical layout capacity. OpenZFS also keeps a slop-space reserve (by default 1/32 of the pool, about 3%) and every filesystem uses some space for metadata, so expect slightly less in practice. SHR results are estimates; the vendor's own tool is the reference for its devices.

Is RAID a backup? ▼

No. RAID keeps an array online when a drive fails, but it copies mistakes just as faithfully: a deleted folder, ransomware encryption, filesystem corruption, a failed controller, a fire or a theft hits every drive at once. A common rule is 3-2-1: three copies of your data, on two kinds of media, with one copy off-site. Test a restore now and then.