Instead of being a SQL Server capability, though,RAID is a feature of the hardware that powers the server. RAID stands for Redundant Array of Independent Disks, and it’s configured by the operating system. RAID arrays use multiple physical hard drives, provide increased disk read and write speeds, more data redundancy protection, or both.
RAID 5 uses disk striping as in RAID 1,but adds on an additional data redundancy protection called parody. With this method, data files are divided across multiple disks, and then an additional physical drive is used to store a full redundant copy of the file. If any single drive fails, the files can be rebuilt either from their striped components or from the parody copy.
RAID level 10, or 1+0, provides the most benefit in terms of performance and fault tolerance, at the expense of using twice the number of disks. With RAID 10, multiple disks are used to create a striped array, just as in RAID 0. Then, the entire striped array is mirrored to a duplicate set of disks as in RAID 1. With this configuration, data files are read and written quickly,since each drive in the striped array contributes to the operation. And, the data’s protected, since each drive is cloned. In a RAID 10 array, as many as half of the disks can fail before any data loss will be experienced.