RAID (Redundant Array of Independent Disks) is a technology that combines multiple physical drives into a single logical storage system to improve fault tolerance, performance, or a combination of both. Data is distributed across the drives according to a specific scheme determined by the selected RAID level.
RAID is used in servers, storage systems, NAS devices, and other IT infrastructure. With RAID levels that provide redundancy, the failure of one or more drives does not necessarily result in data loss or an immediate system outage.
How RAID Works
A RAID controller or software mechanism presents multiple drives to the operating system as a single logical device. Data can be distributed across the drives using three main techniques:
- Striping — data is divided into blocks and distributed across multiple drives. This allows operations to be performed in parallel and can improve performance.
- Mirroring — identical data is written to multiple drives. If one drive fails, another copy remains available.
- Parity — additional information is stored along with the data, allowing the contents of a failed drive to be reconstructed.
Different RAID levels use one or more of these techniques.
Main RAID Levels
RAID 0
RAID 0 uses data striping across two or more drives without providing redundancy.
Advantages:
- high read and write performance
- use of virtually the entire combined capacity of the drives
The main disadvantage is the lack of fault tolerance. Failure of a single drive can make the data across the entire array unavailable.
RAID 1
RAID 1 is based on mirroring. Identical data is written to at least two drives.
Such an array can continue operating if one drive in a mirrored pair fails. However, additional capacity is required to store the copy: with two identical drives, usable capacity is typically equal to the capacity of one drive.
RAID 5
RAID 5 uses data striping with distributed parity. At least three drives are required to create the array.
It can preserve data if one drive fails. When drives of equal capacity are used, usable capacity is approximately equal to the combined capacity of all drives minus the capacity of one drive.
RAID 6
RAID 6 is similar to RAID 5 but uses dual distributed parity. At least four drives are required.
This RAID level can withstand the simultaneous failure of two drives. The additional protection comes at the cost of lower usable capacity and additional overhead during write operations.
RAID 10
RAID 10 (RAID 1+0) combines mirroring and striping. It typically requires at least four drives.
Data is protected through mirroring and then striped across the mirrored groups. RAID 10 provides high performance and fault tolerance, but a significant portion of the total disk capacity is used to store mirrored copies.
Differences Between the Main RAID Levels
The key characteristics of the most common RAID configurations can be summarized as follows:
- RAID 0 — high performance but no protection against drive failure; minimum 2 drives.
- RAID 1 — mirroring and protection against the failure of one drive in a mirrored pair; minimum 2 drives.
- RAID 5 — distributed parity and tolerance of one drive failure; minimum 3 drives.
- RAID 6 — dual parity and tolerance of two drive failures; minimum 4 drives.
- RAID 10 — a combination of mirroring and striping; minimum 4 drives.
The choice of RAID level depends on performance requirements, usable capacity, storage costs, and the number of drive failures the system must be able to tolerate.
Hardware and Software RAID
Based on the implementation method, RAID can be divided into two main types:
- Hardware RAID is managed by a dedicated controller. It handles data distribution, array health monitoring, and recovery after failures. Some controllers include their own memory and cache protection mechanisms.
- Software RAID is implemented by the operating system or other software and uses the server’s computing resources.
Both approaches can be used in enterprise infrastructure. The choice depends on performance, cost, functionality, management, and system architecture requirements.
What Happens When a Drive Fails
The behavior of the array depends on the RAID level. In configurations with redundancy, failure of an allowable number of drives puts the array into a degraded state. Data remains available, but the level of protection is reduced, and performance may also decrease in some cases.
After the failed drive is replaced, a rebuild is performed to reconstruct the data on the new drive using the mirrored copy or parity information.
During a rebuild, the array is under additional load. If more drives fail before the rebuild is complete than the specific RAID level can tolerate, data may become unavailable or be lost.
RAID and Backup
RAID is not a backup. Its primary purpose is to make a storage system resilient to certain drive failures and, depending on the configuration, improve performance.
RAID does not protect against many other threats, including:
- accidental file deletion
- data corruption caused by an application
- malware
- administrator errors
- failure of the entire array or controller
- fire, flooding, and other disasters
- loss of equipment
If a user deletes a file, the change is applied to the array regardless of redundancy. Therefore, RAID should be used together with backups and, where necessary, other data protection and disaster recovery mechanisms.
How to Choose a RAID Level
The choice depends on the purpose of the system and data requirements. Factors to consider include:
- required read and write performance
- the acceptable number of simultaneous drive failures
- required usable storage capacity
- the number and capacity of drives
- array rebuild time
- workload characteristics
- data availability requirements
For example, RAID 0 may be used for workloads where performance is critical and data loss is acceptable. RAID 5 or RAID 6 can be used in systems that require a balance between capacity and fault tolerance. RAID 10 is commonly used in scenarios where both performance and resilience to failures are important.