RAID 5 vs RAID 10
RAID 5 gives you more usable capacity. RAID 10 gives you faster writes and far safer rebuilds. For modern large drives and any write-heavy workload, RAID 10 is the safer call, and on big disks RAID 5 is now actively discouraged. The one case for RAID 5 is squeezing maximum capacity out of a small, read-heavy array. Here is the full picture.
Updated 2026-06-01 · by Jonathan Caruso
Side by side
| RAID 5 | RAID 10 | |
|---|---|---|
| How it works | Striping with single distributed parity | Striped pairs of mirrors |
| Minimum drives | 3 | 4 (even count) |
| Usable capacity | n - 1 drives | n / 2 drives (50%) |
| Drives that can fail | 1 | At least 1 (up to one per mirror) |
| Read performance | Good | Very good |
| Write performance | Slower (parity calculation) | Fast |
| Rebuild | Slow, stresses every disk | Fast, copies one mirror |
| Risk on large modern drives | High | Low |
How each one works
RAID 5 stripes data across all drives and spreads one drive's worth of parity across the set. With three or more disks you lose the capacity of one to parity, and the array survives a single drive failure by rebuilding the missing data from parity. It is capacity-efficient, because only one drive is spent on redundancy no matter how many you have.
RAID 10 is a stripe of mirrors. Drives are paired into mirrors, and those mirrors are striped together for speed. You lose half your raw capacity, because every drive has a full copy in its mirror, but you get fast reads and writes and a simple, low-stress recovery: replacing a failed drive just copies its mirror, with no parity math.
The RAID 5 rebuild problem
This is the reason storage people steer you away from RAID 5 on big disks. When a drive fails, RAID 5 rebuilds the replacement by reading every block on every remaining drive to recompute parity. On modern 8 TB and larger drives, that rebuild takes many hours to days, and it hammers the surviving disks the whole time, which are the same age and have the same wear.
Two things can go wrong during that window. A second drive can fail outright, and with single parity that loses the entire array. Or one of the surviving drives can hit an unrecoverable read error on a single sector, which during a RAID 5 rebuild can also fail the rebuild. The bigger the drives, the higher the odds, which is why RAID 6 (two parity drives) or RAID 10 has largely replaced RAID 5 for anything that matters.
Performance and capacity, the real tradeoff
RAID 5 wins on usable capacity. With four 8 TB drives you get about 24 TB usable versus 16 TB on RAID 10. If your workload is mostly reads and capacity per dollar is the goal, that efficiency is real.
RAID 10 wins on write performance and recovery. There is no parity to calculate on every write, so write-heavy workloads like databases, busy VMs, and lots of small files run noticeably faster. Rebuilds are fast and low-stress because they copy a single mirror rather than reading the whole array. For anything write-heavy or latency-sensitive, RAID 10 is the better engine.
What to actually run on a homelab
For a write-heavy box (VM storage, databases, an active homelab), run RAID 10. For a large, mostly-read media array where capacity matters most, do not reach for RAID 5. Use RAID 6 or a ZFS RAIDZ2 vdev instead, which adds a second parity drive and removes the single-failure-during-rebuild trap while keeping good capacity efficiency.
If you run ZFS (TrueNAS), the equivalents are a stripe of mirrors (the RAID 10 analog) versus RAIDZ1 (RAID 5) and RAIDZ2 (RAID 6). The same advice applies: mirrors for speed and easy expansion, RAIDZ2 for capacity with safety, and RAIDZ1 only on small or non-critical pools.
Before you buy drives, run your exact disk count and layout through the RAID and ZFS storage calculator on this site. It shows usable capacity, the split between data and parity, efficiency, and how many drives can fail, so you can compare RAID 5, RAID 6, and RAID 10 on your real numbers.
Where RAID 5 wins
- Best usable capacity: only one drive is spent on parity no matter how many you have.
- Good read performance for media and file serving.
- Works with as few as three drives.
Where RAID 10 wins
- Fast reads and writes with no parity penalty.
- Fast, low-stress rebuilds that copy one mirror instead of reading the whole array.
- Far safer on large modern drives, where RAID 5 rebuilds are risky.
Which to pick, by situation
| Your situation | Pick | Why |
|---|---|---|
| VM storage or databases (write-heavy) | RAID 10 | No parity penalty on writes, and fast low-stress rebuilds. |
| Large drives, 8 TB and up | RAID 10 | RAID 5 rebuilds are too risky on big disks. Use RAID 10 or RAID 6. |
| Maximize usable capacity, mostly reads | RAID 5 | Only one drive lost to parity, but consider RAID 6 for safety. |
| Four drives, want speed plus redundancy | RAID 10 | Two mirrors striped gives fast reads, writes, and quick recovery. |
The verdict
For a homelab, RAID 10 is the safer default. It is faster on writes, recovers quickly, and avoids the rebuild risk that makes RAID 5 dangerous on large drives. Choose RAID 5 only when you need maximum usable capacity on a small, mostly-read array, and even then look hard at RAID 6 first. Run your real disk count through the storage calculator before deciding, so you are comparing actual usable capacity and fault tolerance, not rules of thumb. If you are picking a NAS OS to run it on, see TrueNAS vs Unraid.
Choose RAID 5 if you want maximum usable capacity on a small, read-heavy array and accept slower writes and rebuilds (and you have considered RAID 6).
Choose RAID 10 if you want fast writes, quick low-stress rebuilds, and safety on large drives, and you can spend half your capacity on redundancy.
FAQ
Should I use RAID 5 or RAID 10?
Use RAID 10 for write-heavy workloads, large drives, or anything where rebuild safety matters. Use RAID 5 only for capacity-efficient, mostly-read arrays on smaller drives, and even then RAID 6 is usually the better choice.
Why should RAID 5 no longer be used?
On large modern drives, a RAID 5 rebuild reads every remaining disk for hours or days, and a second drive failure or an unrecoverable read error during that window loses the whole array. With single parity the margin is too thin, so RAID 6 or RAID 10 has replaced it for important data.
What are the disadvantages of RAID 10?
You lose half your raw capacity, because every drive is mirrored. It also needs an even number of drives, at least four. The tradeoff buys you speed and safe, fast rebuilds.
Does RAID 10 require 4 drives?
Yes, a minimum of four, and an even number beyond that, because it is built from mirrored pairs that are then striped. Two drives would just be RAID 1.
Is RAID 10 the fastest?
For most real workloads, yes, among the redundant levels. It has no parity to calculate, so writes are fast, and striping across mirrors gives strong read speed. RAID 0 is faster but has no redundancy at all.
What is the difference between RAID 0, 1, 5, 6, and 10?
RAID 0 stripes for speed with no redundancy. RAID 1 mirrors two drives. RAID 5 stripes with one parity drive (survives one failure). RAID 6 stripes with two parity drives (survives two). RAID 10 stripes mirrored pairs for speed plus redundancy. Our storage calculator shows usable capacity and fault tolerance for each.