RAID (Redundant Array of Independent Disks) is the technique of creating a large reliable storage solution using multiple computer HDDs or SDDs for better recovery plan. Based upon the RAID type implemented, different techniques like striping, mirroring and/or parity is used to achieve the result.
Common RAID types
This technique strips data across two or more disks in chunks of stripe size which is defined at the time of creation. Since it doesn’t implements parity and mirroring across the disks, it doesn’t provide fault tolerance and redundancy. Use of unequal disk sizes will result on use of smallest disk size on each disk.
RAID 1 mirrors the data between two or more disks without parity or striping. As a result, it is doesn’t provide fault tolerance or spanning of disks. Since an exact data is replicated over all the disks, the disk capacity limit is restricted to smallest disk size in the array. The same is true for the disk performance.
This technique uses at-least 3 disks to achieve block-level stripping (each block defaults to 512 bytes but can be changed during RAID creation) with distributed parity. Here the parity sector is evenly distributed across each disks and failure of single disk won’t cause data loss.
This technique combines two or more standard RAIDs to achieve better performance and/or redundancy. It usually denotes combined numeric representation of standard RAID level used in the name with first numeric denotation implying lowest RAID level with rightmost number representing highest RAID used.
Here data stripes written first and mirrored to another set of disks thereby requiring atleast 4 disks.
This technique first mirrors the data block to a disk and continues to process for each data stripe om available disk pairs. This technique also requires minimum 4 disks.
Other examples of hybrid RAIDs are RAID 50, RAID 100
For more details, visit the wikipedia page.