Key Points
- RAID storage combines multiple drives into one logical array to improve performance and/or provide fault tolerance.
- RAID 0 uses disk striping for faster performance but provides no redundancy, so one failed drive can cause the entire array to fail.
- RAID 1 mirrors data across at least two drives and keeps it available if one drive in a two-disk mirror fails.
- RAID 5 uses single distributed parity across at least three drives, while RAID 6 uses dual parity across at least four drives to tolerate two drive failures.
- RAID 10 combines disk striping and mirroring to deliver strong read and write performance with 50% usable capacity in a standard two-way mirror configuration.
- Choose a RAID level based on workload, usable capacity, performance, and fault-tolerance needs, and always maintain separate backups because RAID is not a backup.
RAID storage combines multiple storage devices into one logical array to improve performance and/or reliability. Depending on the RAID level, it can improve server performance and help keep critical business data available after a drive failure.
This guide explains everything you need to know about RAID storage, including the different RAID levels (RAID 0, RAID 1, RAID 5, RAID 6, and RAID 10), what each of them does, and their advantages and use cases. This information will help you choose the right level for your specific needs, from home NAS to mission-critical business data applications.
If you prefer a visual overview, be sure to watch our video guide: What Is Redundant Array of Independent Disks (RAID) Storage?
Introduction to RAID storage
RAID (Redundant Array of Independent Disks) storage operates two or more disks in tandem, spreading or duplicating (or both) data across them. RAID has been around for a long time and is the cornerstone of high-performance, high-reliability data storage.
RAID is a widely supported standard — from consumer PC motherboards to off-the-shelf network-attached storage (NAS) solutions, you can implement RAID for your home or small business, and, depending on the RAID level, it can help keep data available after a drive failure. However, RAID does not replace a broader backup strategy.
For larger organizations and those with high-performance computing requirements, RAID arrays can provide huge performance increases (especially when using SSDs) as well as data redundancy, ensuring mission-critical workloads are completed quickly and that both workflows and their data can be quickly recovered in the event of hardware failure.
Understanding RAID levels (RAID 0, RAID 1, RAID 5, RAID 6, RAID 10) and how they work
The hardware or software that manages a RAID configuration is called a RAID controller. RAID controllers provide several widely adopted configuration standards known as levels. Each level is optimized for a specific scenario focusing on performance, redundancy, fault tolerance, or a mix, depending on the number of physical storage devices available.
Before choosing which RAID level is right for your system, you need to figure out which metric is most important and your budget. If performance is your primary concern and the data can be recreated, you may choose RAID 0. If you need basic drive redundancy, RAID 1 may be suitable for a system with two drives. If you need high performance as well as the highest level of data protection for valuable data, you’ll need specialized hard disks designed for continuous operation in RAID deployments and more of them.
Each RAID level requires different hardware and software configurations and has different advantages. The most commonly used RAID levels are explained below.
RAID 0 (striping)
RAID 0 stripes (distributes) data across all disks in the RAID array. It requires at least 2 two disks.
| Performance | Redundancy | Storage Efficiency |
| High read and write speeds. The more disks, the faster the RAID 0 array. | None. A single disk failure makes the array unusable. | 100% as all storage space is used for a single copy of data. |
The ideal use case for RAID 0 is high-performance applications where losing data isn’t critical. This might include scratch disks for media editing or for high-performance workloads on data that has been staged from highly redundant (but slower) storage elsewhere.
RAID 1 (mirroring)
RAID 1 stores an exact copy of your data on each disk in a RAID array, requiring at least two disks.
| Performance | Redundancy | Storage Efficiency |
| Reading data is faster (as different parts of data can be read from multiple disks simultaneously), but writing data is slower as it must be written to every disk at the same time. | High. Data can be recovered if a disk fails. The more disks you have in RAID1, the higher the redundancy. | Since the same data is stored on each disk, efficiency decreases with each additional disk added (i.e., a RAID 1 array with two disks has 50% storage efficiency, a RAID 1 array with three disks has 33.33% efficiency, etc.). |
RAID 1 is often used for mission-critical data where availability is a priority and a single drive failure must not interrupt access. Separate backups are still needed to protect against other causes of data loss. This includes core business data (like financial records) and data required for business continuity. Individuals also build RAID 1 arrays for storing their important family photos and documents so that a single drive failure does not wipe out their precious memories. A separate backup is still needed in case the entire device is lost or damaged.
RAID 5 (striping with parity)
RAID 5 adds protection with distributed parity. Parity information is calculated from the stored data and can be used to reconstruct missing data if one disk fails. Parity data is spread across all disks in a RAID 5 array, allowing data to be reconstructed in the event of a single disk failure, but requires additional disks (at least three) due to the need to store the parity data for each disk.
| Performance | Redundancy | Storage Efficiency |
| Read performance is improved as data can be read from disks simultaneously, but write speed is decreased due to the need to calculate and write parity data. For this reason, RAID 5 is often implemented using dedicated hardware to speed up parity calculations. | If a single disk fails, the data on it can be rebuilt using the parity data stored about it on other disks in the RAID 5 array. | One disk’s total storage space is used for parity data, spread across all disks in the array. |
RAID 5 is commonly deployed on servers where both cost and fault tolerance are factors, as it provides a good compromise among storage efficiency, performance, and redundancy.
RAID 6 (striping with double parity)
RAID 6 works the same way as RAID 5, but uses two independent sets of parity information distributed across disks. This allows the array to tolerate up to two disk failures and still rebuild the missing data. RAID 6 requires at least four disks.
| Performance | Redundancy | Storage Efficiency |
| Faster read performance due to multiple simultaneous reads, but slower write performance due to multiple parity calculations. | High, two concurrent disk failures can occur with no data loss. | Two disks worth of storage space are lost to parity data. |
RAID 6 is popular in situations where RAID 5 does not guarantee enough fault tolerance.
RAID 10 (combined mirroring and striping)
RAID 10 (1 + 0) is a nested RAID configuration consisting of RAID 1 and RAID 0 in combination. RAID 10 combines mirrored RAID 1 arrays and stripes them in RAID 0, resulting in high speed and redundancy. You can also use RAID 01 (with the mirroring and striping reversed from RAID 10), which offers less fault tolerance with the same storage requirements; it is not recommended.
| Performance | Redundancy | Storage Efficiency |
| High read and write performance, comparable to a RAID 0 configuration. | High, due to mirroring. Raid 10 can tolerate one failed disk in each mirrored pair, but the array fails if both disks in the same two-disk mirror fail. | At least four disks are required for RAID 10, and only 50% of the total storage space is available as the rest is taken up by mirrored data (assuming two mirrors are present). |
Other RAID levels can also be combined or nested. For example, RAID 50 combines multiple RAID 5 arrays and stripes data across them using RAID 0. This provides the fault tolerance of RAID 5 while improving performance through striping. RAID 50 requires a minimum of six physical storage devices.
As RAID configurations aren’t standardized by a central authority, the details of implementing the above can vary from vendor to vendor, so you should always check what RAID features your hardware and software support, and that they meet your technical requirements. Additionally, there are some less-often seen RAID levels, such as RAID 2, 3, and 4, that implement different striping methods but do not offer any practical advantages over the standard RAID levels.
See how centralized monitoring and alerts help IT teams identify infrastructure problems sooner
Choosing and implementing the right RAID storage
There are several ways that you can deploy RAID storage:
- Software RAID: Linux supports software RAID through the MD subsystem. Windows Server provides RAID-like simple, mirror, and parity resiliency through Storage Spaces, while macOS Disk Utility can create striped and mirrored RAID sets. Available features and supported layouts vary by operating system. Third-party RAID software is also available.
- Hardware RAID: Dedicated RAID hardware (usually connected via PCI Express for use in servers) can provide high storage performance and offload RAID processing from the host system, for a price. Many consumer motherboards also ship with RAID configuration utilities but may differ in performance and features from dedicated RAID hardware.
- Network attached storage (NAS): Many consumer and professional NAS devices offer RAID out of the box — just add your own drives, set the RAID level, and you’re set.
Test monitoring and alerting workflows across your IT environment
Managing resiliency and integrity in RAID for mission-critical data
RAID storage is not a panacea for protecting against data loss. Even the most extensive RAID array can lose data due to accidental deletion, ransomware, file system corruption, misconfiguration, controller failure, theft, fire, or a natural disaster.
A common failure in RAID storage systems is simply that nobody notices a disk failure, so it’s important to monitor your RAID storage as part of a unified endpoint management platform and to keep backups or redundant copies of your data stored off-site for disaster recovery.
If you’re planning to set up a data center, watch our video: JBOD vs RAID: Which One Is Best for Data Centers? to find out which setup is best for you.
