RAID levels define how multiple drives combine to balance speed, redundancy, and usable capacity in a single storage array. RAID, short for redundant array of independent disks, uses striping, mirroring, and parity to spread or duplicate data across drives.
Each RAID level sets a different tradeoff between performance, fault tolerance, and the minimum number of drives required. This guide defines RAID, explains striping, mirroring, and parity, details RAID 0, 1, 5, 6, and 10, gives the usable-capacity formulas, compares hardware and software RAID, and clarifies why RAID is not a backup.
What Is RAID?
RAID is a storage technology that combines multiple physical drives into a single logical unit for performance, redundancy, or both. RAID (redundant array of independent disks) distributes data across two or more drives using striping, mirroring, or parity to improve speed, protect against drive failure, or both. The RAID concept originated in a 1987 University of California, Berkeley paper that defined the original levels.
A RAID controller or software layer presents the array to the operating system as one volume. The role of RAID within the broader secondary storage layer is to raise availability and throughput beyond what a single drive provides.
What Are Striping, Mirroring, and Parity?
Striping, mirroring, and parity are the three data-distribution methods that define every RAID level. Striping splits data across drives for speed, mirroring duplicates data for redundancy, and parity stores calculated recovery information that rebuilds data after a failure. Striping writes consecutive blocks to different drives so reads and writes run in parallel.
Mirroring writes identical copies to two or more drives so a surviving copy preserves the data. Parity uses an XOR calculation to store recovery data that reconstructs a failed drive’s contents from the remaining drives.
The three methods below form the building blocks of RAID:
- Striping spreads blocks across drives to multiply read and write throughput, with no redundancy on its own.
- Mirroring copies data to a second drive so the array survives the loss of one drive in a pair.
- Parity calculates recovery data with an XOR operation, storing enough information to rebuild one or two failed drives.
What Is RAID 0?
RAID 0 stripes data across drives for maximum speed with no redundancy. RAID 0 requires a minimum of 2 drives, stripes data across all of them, provides 100% usable capacity, and offers no fault tolerance, so any single drive failure loses the entire array. Two drives in RAID 0 roughly double sequential read and write throughput because the controller reads and writes both drives in parallel.
RAID 0 suits scratch disks, video editing caches, and temporary data where speed matters and the content is reproducible. The combined failure risk rises with each added drive because a single failure destroys all data on the array.
What Is RAID 1?
RAID 1 mirrors identical data across drives for redundancy. RAID 1 requires a minimum of 2 drives, writes the same data to each, provides 50% usable capacity, and tolerates the failure of all but one drive in the mirror set. A two-drive RAID 1 array survives one drive failure with no data loss because the surviving drive holds a complete copy.
Read performance can improve because the controller reads from either drive, while write performance matches a single drive. RAID 1 suits operating system drives, small business servers, and any data requiring continuous availability with a simple two-drive setup.
What Is RAID 5?
RAID 5 stripes data with distributed single parity for a balance of capacity and redundancy. RAID 5 requires a minimum of 3 drives, stripes data and parity across all drives, provides usable capacity of (n-1) drives, and tolerates a single drive failure. A four-drive RAID 5 array gives three drives of usable capacity and survives one failed drive, rebuilding the lost data from parity.
Write performance carries a parity-calculation penalty, and a rebuild stresses the remaining drives, raising the risk of a second failure during recovery on large arrays. RAID 5 suits file servers and NAS units that need capacity efficiency with single-drive protection.
What Is RAID 6?
RAID 6 extends RAID 5 with double parity for tolerance of two drive failures. RAID 6 requires a minimum of 4 drives, stripes data with two independent parity blocks, provides usable capacity of (n-2) drives, and tolerates the simultaneous failure of any two drives. A six-drive RAID 6 array gives four drives of usable capacity and survives two failures, which protects against a second drive failing during a long rebuild.
The second parity block adds more write overhead than RAID 5. RAID 6 suits large arrays of high-capacity drives where rebuild times are long and the chance of a concurrent second failure is significant.
What Is RAID 10?
RAID 10 combines mirroring and striping for both speed and redundancy. RAID 10 requires a minimum of 4 drives, mirrors pairs of drives and stripes across the mirrors, provides 50% usable capacity, and tolerates one drive failure per mirror set. A four-drive RAID 10 array stripes across two mirrored pairs, delivering high read and write speed while surviving a single failure in each pair.
Rebuilds are fast because the array copies from the surviving mirror rather than recalculating parity. RAID 10 suits databases and transactional workloads that need both high performance and strong fault tolerance, accepting the 50% capacity cost.
How Do You Calculate Usable Capacity?
Usable capacity follows a fixed formula for each RAID level based on drive count. RAID 0 yields full capacity (n drives), RAID 1 yields half (n/2 for a pair), RAID 5 yields (n-1) drives, RAID 6 yields (n-2) drives, and RAID 10 yields half (n/2) of the total raw capacity. Arrays assume equal-size drives, and a mixed-size array limits each drive to the smallest member’s capacity.
Four 4 TB drives produce 16 TB in RAID 0, 8 TB in RAID 10, 12 TB in RAID 5, and 8 TB in RAID 6. The capacity formula always trades usable space for the redundancy the level provides.
The formulas below convert drive count into usable capacity:
- RAID 0 gives n drives of usable capacity, the full raw total with no redundancy reserved.
- RAID 1 gives n/2 drives, half the raw capacity because every block is duplicated.
- RAID 5 gives n-1 drives, reserving one drive’s worth of space for distributed parity.
- RAID 6 gives n-2 drives, reserving two drives’ worth for double parity.
- RAID 10 gives n/2 drives, half the raw capacity from mirroring within striped pairs.
What Is the Difference Between Hardware and Software RAID?
Hardware and software RAID differ in where the array logic runs. Hardware RAID uses a dedicated controller card with its own processor and cache, while software RAID runs the array logic on the host CPU through the operating system. Hardware RAID controllers from Broadcom and Adaptec offload parity calculation and often include battery-backed cache for write protection. Software RAID, such as Linux mdadm, Windows Storage Spaces, and ZFS, uses no extra hardware but consumes host CPU cycles for parity.
Hardware RAID can be faster for parity levels, while software RAID offers flexibility and lower cost. Both present the array as a single volume to the operating system.
Why Is RAID Not a Backup?
RAID is not a backup because it protects against drive failure, not data loss. RAID maintains availability when a drive fails, but it does not protect against accidental deletion, file corruption, ransomware, or the failure of the array controller, since those errors propagate across every drive instantly. A deleted file is deleted on all mirrored or parity drives at once. A backup keeps an independent, point-in-time copy on separate media, which RAID does not provide.
The correct approach uses RAID for uptime and a separate backup strategy for data protection. Choosing the right array also depends on selecting the underlying drives for endurance and reliability.
Key Takeaways
The points below summarize RAID levels:
- RAID combines drives using striping, mirroring, and parity to balance speed, redundancy, and capacity.
- RAID 0 maximizes speed with no redundancy, needing 2 drives and risking total loss on one failure.
- RAID 1 and RAID 10 mirror data, giving 50% usable capacity and fast, simple redundancy.
- RAID 5 and RAID 6 use parity, tolerating one and two drive failures with (n-1) and (n-2) capacity.
- Hardware RAID offloads to a controller, while software RAID runs on the host CPU at lower cost.
- RAID is not a backup, because deletion, corruption, and ransomware propagate across every drive at once.
RAID Levels Comparison
The table compares the five common RAID levels across the attributes that drive selection:
| RAID Level | Minimum Drives | Redundancy | Usable Capacity | Primary Use |
|---|---|---|---|---|
| RAID 0 | 2 | None | 100% (n drives) | Speed, scratch and cache disks |
| RAID 1 | 2 | 1 drive (mirror) | 50% (n/2) | OS drives, small servers |
| RAID 5 | 3 | 1 drive (parity) | (n-1) drives | File servers and NAS |
| RAID 6 | 4 | 2 drives (double parity) | (n-2) drives | Large high-capacity arrays |
| RAID 10 | 4 | 1 per mirror set | 50% (n/2) | Databases and transactional loads |
What is the best RAID level?
No single RAID level is best. RAID 10 balances speed and redundancy for databases, RAID 5 and RAID 6 favor capacity efficiency for file servers, and RAID 0 maximizes speed without protection.
How many drives does RAID 5 need?
RAID 5 requires a minimum of 3 drives. It stripes data and distributed parity across all drives, provides usable capacity of (n-1) drives, and tolerates a single drive failure.
What is the difference between RAID 5 and RAID 6?
RAID 5 uses single parity and tolerates one drive failure with a 3-drive minimum. RAID 6 uses double parity, tolerates two simultaneous failures, and requires a minimum of 4 drives.
Is RAID a backup?
RAID is not a backup. RAID protects against drive failure but not deletion, corruption, or ransomware, which propagate across every drive. A separate point-in-time backup is required for data protection.
What is the difference between RAID 1 and RAID 10?
RAID 1 mirrors data across 2 drives. RAID 10 mirrors pairs and stripes across them with a 4-drive minimum, adding striping speed on top of mirror redundancy.
How much usable capacity does RAID 6 have?
RAID 6 provides usable capacity of (n-2) drives because two drives’ worth of space holds double parity. Six 4 TB drives yield 16 TB usable while tolerating two failures.
Last Thoughts on RAID Levels
RAID levels balance speed, redundancy, and usable capacity across multiple drives. RAID 0 maximizes speed, RAID 1 and RAID 10 mirror for simple redundancy, and RAID 5 and RAID 6 use parity to tolerate one or two drive failures with better capacity efficiency. The minimum drive count, fault tolerance, and capacity formula together guide the right level for a workload.
Because RAID protects uptime rather than data, a separate backup remains essential. Readers can continue with how to choose a storage drive, compare HDD and SSD options, and use the computer hardware guide as the central reference.
