Innovative memory and storage solutions direct from Micron for 45 years.

A RAID is a data storage device. RAID is an acronym that stands for Redundant Array of Independent Disks, or Redundant Array of Inexpensive Disks. 

What are RAID arrays?

A RAID array is a collection of SSDs or hard drives which have been configured to act as one large storage ‘pool’. This makes it possible to build storage systems far bigger than any single hard drive.   

The drives that make up the RAID array can be configured in different ways which means systems can be optimized for reliability or performance. RAID can be hardware or software-based. 

Hardware RAID

Hardware RAID uses a dedicated RAID controller to manage the installed drives. This could be a separate device in the computer, but they can be built into motherboards. A RAID system that uses a hardware controller is entirely dependent on the controller to manage the data flow and storage with the RAID. If the RAID controller becomes inoperative, it may not be possible to recover the RAID array, and you could lose data.  

Investing in dedicated RAID hardware controllers provides better performance than software RAID because they relieve the host system of the work, but they generally cost more.  

Software RAID

Software RAID uses software within your installed operating system to configure and manage the RAID. Software RAID is generally cheaper to configure and use than hardware RAID. Steps to configure RAID vary depending on the exact type being set up, but generally, an array is configured in a RAID adapter's firmware utility or your system's UEFI or BIOS. After that, an operating system looks at the array as a destination to partition and begins writing data to either installing an operating system to the RAID or using it as a secondary volume. Refer to support for your motherboard, operating system, or dedicated RAID adapter for detailed instructions on setup and management. 

RAID for performance

A RAID configuration can be used to optimize storage performance by spreading the data across multiple drives in the array in a process known as striping. When data is saved to a single drive, it has to be written to the device in series, or one bit at a time. Hard drives have a finite limit to how quickly they can read and write data. Bit 1 has to be written first, then bit 2, then bit 3, and so on until the data is complete.   

In a RAID array, this data can be written in parallel across more than one drive, depending on the configuration. Bit 1 can be written to drive 1, bit 2 can be written to drive 2, and bit 3 can be written to drive 3, and so on. Each drive only stores a fragment of the overall data, which means the total write time is reduced. The same is true for data being read. In this way the speed limit of each individual drive is shared, speeding up the operation.  

As each drive now only contains a part of the overall file, all drives need to be operational in order to access the data reliably.  

RAID for reliability

RAIDs can be configured to improve reliability by using a process of mirroring. In this configuration, the data is saved to multiple drives at once. This increases write time, as the data needs to be saved more than once, but it means that the failure of a single drive will not cause any data loss.  

These arrays work best when all connected drives are identical, but in many RAID environments, different disks can be used. Performance and capacity differences across connected drives reduce performance and usable capacity in every disk in the array to that of the lowest performing part.  

RAID as backup

RAID arrays can work as an important part of your overall backup strategy, but they should not be used exclusively. RAIDs can still fail, and as they are typically designed to hold large amounts of data, this loss can be catastrophic. RAID should be used as part of your 3-2-1 backup strategy: 3 copies of your data, in at least 2 locations, and 1 off-site. 

Types of RAID configuration

The notion of using RAID for performance or reliability is a general principle, but RAID arrays can be configured in several ways. Some technology providers have developed proprietary versions of RAID, but the configurations below are the common industry standards. 

RAID 0

This treats all the drives as one large storage volume. This provides the maximum capacity possible, but there is no redundancy provided. The loss of any single drive could cause the entire volume to be lost. 

RAID 1

This mirrors the data to two or more drives. This provides redundancy because a complete copy of the data will exist on each drive in the set. 

RAID 2

This strips data across multiple drives, with Hamming code for error correction. RAID 2 is obsolete now as modern drives have built-in error correction.  

RAID 3

This is similar in principle to RAID 2 but uses byte-level striping (8 bits) instead of bit-level striping on a dedicated parity disk. Performance limitations make it an unpopular choice for modern storage solutions. 

RAID 4

This operates like RAID 2 and RAID 3, but at block, rather than bit or byte level. It also uses a dedicated parity disk. 

RAID 5

Like RAID 4, this operates at the block level, but the parity data is also distributed across the RAID, rather than having a dedicated disk which could become a bottleneck. RAID 5 can tolerate the loss of a single drive before data loss occurs. 

RAID 6

This is similar in principle to RAID 5 but uses double-parity for additional fault-tolerance. In this array, up to two drives could fail but the array could still continue to function. 

In terms of day-to-day use, RAID works similarly to a single disk, but diagnostic tools read data from a RAID configuration differently than they do a single SSD or hard drive. For example, Crucial Storage Executive is not fully compatible with some RAID controllers and configurations, and specific functions such as SMART reporting or firmware updates may not work in these unsupported environments, requiring the RAID to be temporarily disassembled for updates or troubleshooting of individual drives.  

Also, while modern operating systems and RAID drivers allow trim commands to run on SSDs in RAID, legacy operating systems and drivers may not properly support them, meaning functions such as Garbage Collection become more important for maintaining the highest performance from connected SSDs. 

FAQs