RAID Levels

RAID Levels

There are currently several RAID Levels, but originally there were only five.

The variations of RAID (Redundant Array of Inexpensive/independent Disks) have evolved over time and today there are six RAID levels.

Each RAID level is either used as a fixed configuration or as a combination of multiple levels incorporated into one system solution. Here is a brief RAID levels comparison.

Base RAID Levels

  • RAID 0: Implements “striping” and the level has no data redundancy and no fault tolerance. There is no error detection mechanism either, resulting in the loss of all data on the array in case of failure.

    RAID 0

  • RAID 1: This RAID level introduces mirroring. Data is written the same on two drives or more, producing a mirrored set of data. The read request from the computer’s processor for that data is serviced by any drive that has the data needed. This makes system performance much faster.

    RAID 1

  • RAID 2: This level comprises of bit-level striping with Hamming-code parity dedication. All the spindle rotation on the disk is synchronized. This level is only of historical significance now. It was used in many early machines but is not used by today’s systems.

    RAID 2

  • RAID 3: This level is based on byte-level striping with dedicated parity. As with RAID 2, disk spindle rotation is completely synchronized and the striping of data results in each bit in sequence being on a different drive. Today, RAID 3 is not commonly used.

    RAID 3

  • RAID 4: Consists of block-level striping with dedicated parity. Used previously by NetApp, today it has been more or less replaced by RAID 6.

    RAID 4

  • RAID 5: Like RAID 4, this level also does block-level striping with distributed parity, but unlike RAID 4, the information is distributed throughout the drives. This requires all drives but one to be present in order for an operation to be successful. If a drive fails, subsequent reads can be taken from the parity that is distributed to other drives, which results in no data being lost. At least three disks are required for RAID 5.

    RAID 5

  • RAID 6: Unlike other standard RAID levels, RAID 6 uses double distributed parity with block-level striping which provides needed tolerance for faults on two failed drives. RAID groups can be larger as a result, allowing for high-availability systems and large capacity drives.

    RAID 6


RAID 10, also known as RAID 1&0 or RAID 1+0, is a stripe of mirrors, similar to RAID 0+1 except that it is reversed. According to the storage industry association definition, RAID 10 is a RAID 0 array that contains two-way or three-way mirrors and has a minimum of four drives. It is possible to implement a RAID 10 with only two disks and that configuration is referred to as Linux RAID 10.

RAID 10 is typically used for professional applications in high-end configurations. Another big benefit of RAID 10 is the performance improvement over other levels. According to industry benchmarks and manufacturer specifications, RAID 10 performs better and has better latency than all other RAID levels, with the exception of RAID 0. For database, e-mail and web servers, it is the preferred type of all available RAID levels.

RAID 100

Also called RAID 10+0, this level is a stripe of RAID 10s and is the equivalent to a wider RAID 10 array. Often called plaid RAID level because it is striped two ways.

Other RAID Levels

There are other levels of RAID configurations including RAID 50, RAID 53 and RAID 60. RAID 50 uses straight block-level striping of RAID 5, with the parity of distribution found on RAID 5. In order to use RAID 50, at least six drives are needed. RAID 53 is also known as RAID 30 or 0+3. RAID 60 is a straight block-level striping level (RAID 0) combined with the double parity of RAID 6. This RAID level requires at least eight disks.

RAID levels and configurations continue to evolve and change. RAID levels have come a long way from where they were many years ago when the term RAID was first coined!