What Is RAID?
Author: Domeny.ForSale editorial team Published
In brief: RAID is a way of combining several drives into an array, a shared data storage system. Depending on the level chosen, it can improve performance, allow continued operation after a drive failure or combine these benefits. Not every RAID level protects against failure, and none replaces an independent backup.
How does RAID work, and what is it used for?
RAID stands for Redundant Array of Independent Disks. The operating system usually sees this set of drives as a single logical storage device. Data is distributed across the drives, duplicated or stored alongside information that allows it to be reconstructed.
The RAID level determines how data is stored. The level number is not a quality rating: RAID 10 is not simply a “better version” of RAID 6. Each variant uses drive capacity differently and handles drive failures differently.
Arrays are found in servers, workstations and NAS devices, which provide file storage accessible over a network. One important use is maintaining data availability during a drive failure. Performance, however, also depends on the type of drives, how the array is managed and the operations being performed — using RAID alone does not guarantee a speed increase.
Popular RAID levels: capacity and fault tolerance
The levels most commonly compared are RAID 0, 1, 5, 6 and 10. In the table, N means the number of drives, and S means the capacity of the smallest drive. The figures apply to typical configurations and exclude space used by the file system and metadata.
| Level | Minimum number of drives | Usable capacity | Drive failure tolerance |
|---|---|---|---|
| RAID 0 | 2 | N × S | None |
| RAID 1 | 2 | S with full mirroring | In a two-drive setup: one drive failure |
| RAID 5 | 3 | (N − 1) × S | One drive failure |
| RAID 6 | 4 | (N − 2) × S | Two drive failures |
| RAID 10 | 4 | N/2 × S | Depends on which drives fail |
RAID 0 splits data into chunks and writes them to different drives. It allows their combined capacity to be used and may improve performance, but it stores no redundant data. A single drive failure means losing access to all or part of the array’s data; you should assume that the entire dataset will need to be restored.
RAID 1 writes the same data to the drives that form a mirror. In a typical two-drive configuration, the failure of one drive does not interrupt access to files. The trade-off is that only half the combined capacity is usable.
RAID 5 and RAID 6 use parity, additional mathematical information that enables missing data to be reconstructed. RAID 5 tolerates one drive failure, while RAID 6 tolerates two. Calculating and writing parity can reduce write performance, particularly when performing many small operations.
RAID 10 combines mirrored pairs of drives with data distribution across those pairs. A typical setup requires an even number of drives. It can survive several drive failures if a working drive remains in each pair; losing both drives in one pair causes the array to fail.
How do you choose RAID for a home NAS or a business?
Imagine a two-bay NAS with two 4 TB drives. RAID 1 will provide approximately 4 TB of capacity before system overheads and allow continued operation after one drive fails. RAID 0 will provide approximately 8 TB, but the failure of either drive will put the entire dataset at risk. For family photos, a mirror combined with a separate backup is a more sensible choice; RAID 0 may be considered for working files that can be recreated.
With a larger number of drives, three questions can help you decide:
- Can the data be easily recreated? If not, plan a backup first, then choose a RAID level.
- How long can the service be unavailable? RAID 6’s tolerance of two drive failures provides a greater safety margin during rebuilding than RAID 5.
- Which matters more: capacity or intensive writing? RAID 5 and 6 use space more efficiently than mirrors, while RAID 10 can be advantageous for workloads involving many write operations.
Do not assume that drives of different capacities will simply add up. In conventional arrays, larger drives are usually used only up to the capacity of the smallest drive. Check the documentation for your particular NAS or controller for supported levels, expansion rules and drive compatibility.
What happens after a drive fails?
An array with sufficient redundancy — the extra data needed to survive a failure — can operate in a degraded state. This means losing some or all protection against another failure, often with reduced performance as well. After the failed drive is replaced, rebuilding takes place: the missing data is written to the new drive.
Rebuilding makes intensive use of the remaining drives. The time it takes depends on factors including capacity, workload and the RAID implementation; there is no single reliable figure for all devices. Another failure or unreadable portions of data may jeopardise recovery of the full contents.
For this reason, enable error notifications and do not delay replacing a failed drive. Before removing it, check the bay label in the device’s management interface — accidentally removing a working drive can make the situation worse.
Why does RAID not replace a backup?
RAID primarily protects data availability during certain drive failures. It does not create an independent file history: accidentally deleting a document, overwriting it or having it encrypted by malware can affect the entire array. Theft of the device, fire or damage to the entire NAS also fall outside this protection.
A backup is a separate copy that allows data to be restored. A practical setup is a NAS with RAID 1 plus regular backups to a drive that can be disconnected or to a separate service with file versioning. A copy stored away from the NAS location further reduces the risk of losing everything in a single incident.
Before creating an array, copy your existing files — configuration may erase the drives. Then set a backup schedule and test restoring a few files before entrusting important data to the new setup.
See also: What Is 2FA?
Read also
