The most used and most delicate piece of hardware on any webhosting server is the hard drive, and thus it is the part most likely to fail. Because of this, all our servers have RAID (Redundant Array of Independent Disks) storage devices. On such a device, multiple hard drives (6 hard drives on the RAIDs we use) act as a single storage device. 4 of them are used for storage, 1 is a spare drive and another is a drive for parity (information used to rebuild missing data).
When a file is stored on a RAID, it is divided in equal parts and each part is stored on each of the RAID’s drives. When a file is accessed by the server, all its parts are rejoined. If one of the hard drives fail, the data pieces on the failed drive are not lost, as the parity drive contains information used to rebuild the missing pieces. Also, when a drive fails, the spare drive will become a mirror of the missing drive (a process known as “data reconstruction” that takes only about 30 minutes). The bad hard drive is pulled out and replaced for a new one (and we have spare drives on-site).
Downtime? No downtime. Drives can be replaced and can be reconstructed while the RAID device is still serving and storing files. The unit can continue operating even without one of its hard drives.
What does this mean? No downtime due to hard drive failure, one of the most common causes of downtime in over 90% of all webhosts.