RAID Levels Explained

Category: Dedicated Servers · Updated 2026-10-31

RAID (Redundant Array of Independent Disks) combines multiple physical drives into one logical unit to improve speed, provide redundancy against drive failure, or both. RAID 0 only stripes data for speed with no protection, RAID 1 mirrors data across drives, RAID 5 and 6 add parity so the array survives one or two failed drives, and RAID 10 combines mirroring and striping. The right level depends on whether an application needs more speed, more safety, or a balance of both.

What RAID does

A RAID controller, whether built into the server's hardware or handled by software, treats a group of physical drives as one array. Applications and the operating system see a single volume, while the controller decides how data is actually spread across the underlying disks. The goal is one or both of: faster read and write speeds, and the ability to keep working (or recover quickly) after a drive fails.

RAID is not a backup. It protects against a drive hardware failure, not against accidental deletion, ransomware or a bad application update, since a mistake is written to every drive in the array just as fast as a correct write.

Common RAID levels compared

Level	How it works	Minimum drives	Survives a drive failure?	Usable capacity
RAID 0	Striping, data split across drives	2	No	100% of total
RAID 1	Mirroring, identical copy on each drive	2	Yes, one drive	50% of total
RAID 5	Striping with one parity block	3	Yes, one drive	(n-1)/n of total
RAID 6	Striping with two parity blocks	4	Yes, two drives	(n-2)/n of total
RAID 10	Mirrored pairs, then striped	4	Yes, usually one per mirrored pair	50% of total

Why plain striping (RAID 0) is risky

RAID 0 splits each file across every drive in the array, so reads and writes happen in parallel and finish faster. There is no duplicate data anywhere, so if any single drive in the array fails, the whole array is lost, not just that drive's share. RAID 0 is used only where raw speed matters more than the data itself, such as a scratch disk for temporary rendering files, never for a database or a server's only copy of a website.

Parity-based levels: RAID 5 and RAID 6

Instead of storing a full duplicate of every byte, RAID 5 calculates a parity value from the data on the other drives and stores it alongside the data, spread evenly across all drives. If one drive fails, the controller rebuilds its contents from the parity on the surviving drives. RAID 6 stores a second, independent parity value, so it can lose two drives at once, at the cost of a little more capacity and write overhead. Rebuilding a large, busy RAID 5 array after a failure is itself a slow, disk-intensive process, which is why RAID 6 is often preferred on large arrays: it tolerates a second failure during that rebuild window.

Choosing a level for a hosting server

RAID 1: the usual choice for a server's operating system and boot drive, simple and reliable with just two drives.
RAID 10: a common choice for busy databases, since it mirrors for safety and stripes for speed, at the cost of using only half the raw capacity.
RAID 5 or 6: suited to large file storage or backups where capacity matters more than top write speed.
RAID 0: only for disposable, non-critical, speed-first data.

Whatever level is chosen, RAID only covers drive failure. Off-array, off-site backups are still necessary for deletions, corruption and disasters.

Frequently asked questions

What does RAID stand for?

RAID stands for Redundant Array of Independent Disks. It is a method of combining multiple physical drives into one logical unit for speed, redundancy, or both.

Is RAID a substitute for backups?

No. RAID protects against a physical drive failure, but a deleted file, a corrupted database or ransomware is written to every drive in the array just as reliably as good data, so separate backups are still required.

Which RAID level is best for a database server?

RAID 10 is a common choice for busy databases because it combines mirroring for safety with striping for speed, though it uses only half the raw drive capacity. RAID 1 is a simpler, cheaper option for smaller workloads.

What happens when a drive fails in a RAID array?

In RAID 1, 5, 6 or 10, the array keeps running in a degraded state using redundant data, and the failed drive can usually be swapped and rebuilt without taking the server offline. In RAID 0, a single drive failure causes total data loss for that array.

Can I use RAID with SSDs and NVMe drives?

Yes. RAID works with SSD and NVMe drives the same way it does with spinning hard drives, and many hosting providers now offer RAID 1 or RAID 10 arrays built from SSDs or NVMe for both speed and redundancy.

What is the difference between hardware RAID and software RAID?

Hardware RAID uses a dedicated controller card with its own processor to manage the array, which frees up the server's CPU. Software RAID is managed by the operating system instead, which is cheaper and flexible but uses a small amount of the server's own CPU and memory.