How does the RAID level affect capacity?

Redundant Array of Independent Disks (RAID) technology uses three techniques to improve performance and protect your data: striping, mirroring, and parity.

Striping

Striping breaks down the data into blocks that are written to several separate . The blocks are written to multiple physical disks simultaneously, thereby improving performance. Striping alone does not protect data.

Mirroring

Mirroring provides data protection by writing identical copies of all data to separate physical disks. If one physical disk fails, the system can instantly begin using the physical disk that contains the copy. The cost of mirroring is high because you must use twice as much storage capacity.

Parity

Parity provides data protection by creating redundancy data based on calculations on the data instead of completely duplicating the data. If a physical disk fails, you can use the redundancy data to quickly reconstruct the information. Implementation cost is lower than for mirroring because you do not have to use twice the storage capacity. However, performance is also lower because of the extra processing time needed to calculate the redundancy data.

RAID 0

RAID 0 uses striping to improve performance but provides no data protection. RAID 0 stripes data across all of the physical disks in the . If a single physical disk fails in the disk group, all of the associated fail, and all data is lost.

RAID 0 performance is fast because blocks of data are written to multiple physical disks simultaneously. RAID 0 also provides the most efficient use of storage capacity (the lowest cost) because no storage space is used for mirroring or parity. RAID 0 requires a minimum of two physical disks.

RAID 1 or RAID 10

RAID 1 uses mirroring to write two copies of the data to two different physical disks simultaneously. If a single physical disk fails, the mirror physical disk continues to allow access to the data. If a mirrored pair in a disk group fails, all of the associated virtual disks fail, too.

RAID 1 offers the best data availability. If an error occurs, the storage array can instantly begin using the other physical disk without any loss of data or service. The minimum number of physical disks required is two—one for the user data and one for the mirrored data. If you choose RAID 1 in the MD Storage Manager Software and select four or more physical disks, RAID 10 is automatically configured across the disk group—two physical disks for the user data and two physical disks for the mirrored data.

RAID 5

RAID 5 stripes user data and parity data across all of the physical disks in a disk group, removing the bottleneck of a single parity physical disk. If a single physical disk fails in a RAID 5 disk group, the redundant information allows the data to be reconstructed. If two or more physical disks fail in a RAID 5 disk group, all of the associated virtual disks fail, and all data is lost.

RAID 5 requires a minimum of three physical disks. Using more physical disks provides greater performance and higher storage efficiency.

RAID 6

RAID 6 stripes user data and parity data across all of the physical disks in a disk group, removing the bottleneck of a single parity physical disk. RAID 6 uses the equivalent capacity of two physical disks for redundant information. If two physical disks fail in a RAID 6 disk group, the redundant information allows the data to be reconstructed. If three or more physical disks fail in a RAID 6 disk group, all of the associated virtual disks fail, and all data is lost.

RAID 6 requires a minimum of five physical disks. Using more physical disks provides greater performance and higher storage efficiency.

RAID Level Comparison