keepcalculator

06 Numbers and Powers

RAID 5 vs RAID 10: Speed, Safety & Rebuilds

Card No. 06·12032 · Updated · Standard formula, shown below

The calculator on this card is switched off while it is checked. The formula is below, and the worked example shows how to do the sum by hand.

RAID 5 and RAID 10 are both popular disk arrays, but they answer different needs: RAID 5 maximizes usable capacity with single-drive fault tolerance, while RAID 10 doubles speed and tolerates up to two drive failures but gives you only half the raw capacity. This article explains how each works, their real-world trade-offs, and which one fits common workloads.

Formula

For RAID 5, usable capacity = (number of drives − 1) × capacity of the smallest drive. For RAID 10, usable capacity = (number of drives ÷ 2) × capacity of the smallest drive, assuming you create mirrored pairs. The overhead is the percentage of raw capacity lost to parity or mirroring.

  • RAID 5 overhead = 1 ÷ N (one drive’s worth of parity)
  • RAID 10 overhead = 50% (each drive is mirrored)

Worked Example

Using the defaults: 4 drives of 1000 GB each. For RAID 5: usable = (4 − 1) × 1000 = 3000 GB, overhead = 1 ÷ 4 = 25%. For RAID 10: usable = (4 ÷ 2) × 1000 = 2000 GB, overhead = 50%. The calculator shows 3000 GB for RAID 5 and 2000 GB for RAID 10 with 4 drives.

How RAID 5 Works

RAID 5 stripes data blocks across all drives and writes a parity block on each drive. Parity is a mathematical checksum that lets the array reconstruct data if any single drive fails. With three drives, you get two drives of data and one of parity; with four drives, three of data and one of parity. The parity is distributed, so no single drive is a dedicated parity disk, which avoids the write bottleneck of older RAID 4.

Read performance is excellent because all drives serve data simultaneously. Write performance is lower than a single drive because every write must also update the parity block, which requires reading the old data and parity, computing the new parity, and writing both. That is called the read-modify-write penalty.

How RAID 10 Works

RAID 10 (also called RAID 1+0) first mirrors pairs of drives, then stripes across the mirrored pairs. A four-drive RAID 10 is two mirrored pairs striped together. Every write goes to both drives in a pair, so write speed is similar to a single drive, but reads can be served from either copy, doubling read performance.

Fault tolerance depends on which drives fail. If one drive in each mirrored pair fails, the array survives. If both drives in the same pair fail, the array is lost. With four drives, you can lose up to two drives, but only if they are in different pairs. With six drives, you can lose up to three drives, again one per pair.

Comparing Performance and Safety

For random reads and writes, RAID 10 generally outperforms RAID 5 because it has no parity calculation overhead. For sequential reads, both are fast, but RAID 10 often edges ahead. RAID 5’s write penalty is most noticeable with small random writes, common in database transaction logs and virtual machine disks.

Rebuild time is another major difference. When a drive fails in RAID 5, the array must read all remaining drives to reconstruct the missing data. During that time, performance drops and another failure means total data loss. RAID 10 rebuilds by copying data from the surviving mirror, which is faster and puts less stress on the other drives.

Cost and Capacity Considerations

RAID 5 gives you more usable space for the same number of drives. With four 1 TB drives, RAID 5 offers 3 TB usable, while RAID 10 offers only 2 TB. That extra capacity comes at the cost of lower write performance and longer rebuilds. RAID 10 requires at least four drives and gives you half the raw capacity, but it offers better performance and faster recovery.

If you are building a large media archive where reads dominate and writes are infrequent, RAID 5’s capacity advantage is attractive. If you are running a busy database or virtualization host where write latency matters, RAID 10 is usually the safer choice.

Common Issues and Misconceptions

A common misconception is that RAID 5 can survive two drive failures. It cannot—only one. RAID 6 adds a second parity block and can survive two failures, but it costs even more capacity and write performance.

Another issue is the RAID 5 write hole: if power fails during a write, the parity may be inconsistent, and the array may not detect it until a drive fails. Most modern RAID controllers handle this with battery-backed cache or journaling, but it is still a risk.

Also note that RAID is not a backup. If you accidentally delete a file, or a virus corrupts data, RAID does not help. RAID protects against drive failure, not human error or software corruption. Always keep a separate backup.

How to Choose

Consider your workload and tolerance for risk:

  • Choose RAID 5 if you need maximum usable capacity, have mostly read workloads (file server, media storage), and can accept slower writes and longer rebuilds.
  • Choose RAID 10 if you run databases, email servers, or virtual machines where write speed and quick recovery are critical, and you can afford half the capacity.
  • Choose RAID 6 if you have very large drives (8 TB+) and worry about rebuild times or a second failure during rebuild.

Also check whether your RAID controller supports the level you want. Many motherboard chipsets offer RAID 5, but some entry-level controllers do not. Software RAID (like ZFS or mdadm) can also implement these levels and often provides better flexibility.

Frequently Asked Questions

Can I mix different drive sizes in RAID 5 or RAID 10?

Yes, but the array uses the smallest drive’s capacity for every drive. Extra space on larger drives is wasted. For example, three drives of 1 TB, 2 TB, and 2 TB in RAID 5 give you 2 TB usable (3 × 1 TB − 1 TB), not 3 TB. It is best to use identical drives.

Is RAID 5 safe for SSDs?

Yes, but SSDs have a limited number of writes, and the parity write penalty can wear them faster. RAID 10 with SSDs is often preferred for performance and endurance, but RAID 5 on SSDs works fine for many workloads, especially if the controller supports TRIM passthrough.

What happens if two drives fail in RAID 5?

The array is lost. RAID 5 has only one parity block, so it can tolerate exactly one drive failure. If a second drive fails before the first is rebuilt, all data is unrecoverable without an external backup.

Which is better for a home NAS?

For a home NAS storing media and documents, RAID 5 (or RAID 6 for larger arrays) is popular because it maximizes capacity. If you run VMs or a database on the NAS, RAID 10 gives better performance. Many home users choose RAID 1 (mirror) for simplicity.

Does RAID 10 really double read speed?

Yes, in theory. Because each mirrored pair can serve reads from either drive, concurrent reads can be split across drives. In practice, the gain depends on the controller and workload. Sequential reads may not double, but random reads often see a significant improvement.

Related

Formula used
in n | Number of drives | - | 4 | 3 | 16
in c | Drive capacity | GB | 1000 | 1 | 20000
in t | RAID type | - | RAID 5=5, RAID 10=10 | 5 | 5
out u | Usable capacity | GB | (t == 5 ? (n - 1) * c : (n / 2) * c) + 0 * sqrt(n - (t == 5 ? 2 : 3)) | 0
out f | Drive failures tolerated | - | (t == 5 ? 1 : floor(n / 2)) | 0
out o | Overhead | % | (t == 5 ? (1 / n) * 100 : 50) | 1

Each line is one input (in, sel, list) or one result (out: label, unit, formula, decimals). How the calculators work

Method: the standard formula for this calculation, rounded as each result states. See how the calculators work or report an error.

More cards in this drawer

The whole Numbers and Powers drawer →