RAID stands for ‘Redundant Array of Independent Drives’ (originally it stood for ‘Redundant Array of Inexpensive Disks’ but, due to newly developed Solid-State Drives, this ‘Inexpensive’ term doesn’t hold much weight anymore). This refers to a virtual data storage technology that combines multiple HDD’s or SSD’s into one logical ‘virtual’ unit for the purpose of data redundancy and, in some cases, performance improvement.
The concept of this technology is to provide a level of protection to your data in the server host. At the beginning, it was also meant to save money on larger drives. Instead of getting one big capacity drive, you would get several smaller and cheaper capacity drives and combine them together into one logical drive with the same capacity as the bigger drive but usually for less money (thus, the inexpensive part of the abbreviation). These days, 2TB, 3TB and even 4TB are not that expensive to acquire and today RAID is meant for fault tolerance and performance improvement, rather than to save on cost.
The RAID technology is divided into different types of RAID levels. The level number does not indicate better performance or fault raid controller tolerance, but simply a different numbering. The RAID level preference depends on multiple variables where the server type, RAID controller type, cost of investment, total capacity and future expansion matters. The idea is to find the most suitable protection and performance while accounting for all sorts of hardware variables. Sometimes it’s not that easy to accomplish and requires a lot of calculation.
A very important concept is to understand is that RAID in no way provides any type of backup to the data. If the server sustains physical damage (earthquake, fire, tornado, Sharknado, etc) the data is lost no matter the RAID level. Backup solutions will be reviewed in a future nugget. For complete protection, a system administrator must consider implementing RAID and backup solutions, preferably off-site.
The RAID can be done using a dedicated controller, called the RAID controller, or can be embedded in the operating system. While the later will be cheaper and probably for free (nowadays it comes embedded in the motherboard), the purchase of a dedicated controller is highly recommended as the software RAID is done by the CPU and, upon a failure, rebuilding the RAID will use most of the CPU resource pool for this task, greatly degrading the computer performance for any other tasks. The dedicated controller, though it’s a more expansive option, usually supports more RAID options and can rebuild the RAID in case of a failure in the background. This allows the CPU to focus on what its good at – calculations – and doesn’t stop the user from continued usage of the unit.