What is RAID?
You may have heard about various storage devices being RAID capable, built to work on a RAID, or including RAID right out of the box. These devices are referring to a set of standards that allows multiple independent hard disks to connect to one-another as a single drive. There are a number of different RAID configurations, and in this article, we’ll take a look at some of them and explore when and how they are used.
RAID stands for Redundant Array of Independent Disks, a collection of independent disks that may include two or more individual hard drives that work in a coordinated pattern to provide redundancy, an increase in speed, increased storage capacity, or a mixture of features that improve on the capabilities of a single drive.
RAID 0: Striped Set Without Mirroring or Parity
Perhaps the most basic RAID configuration is RAID 0. This level takes two or more disks and couples them so that information being written is split between the two disks evenly so that every other block of data is sent to the opposing disk. This allows the user to essentially double the read/write speed while creating what the computer sees as a single larger disk. The downside of RAID 0 is a lack of parity. If one disk fails in a RAID 0 setup, all of the data stored in that RAID is lost. RAID 0 is typically used to boost performance, and can be found in external drives intended for use in video editing where speed is absolutely essential. If you have two drives in a RAID 0 configuration of different size, the total space available to the user will equal double the capacity of the smaller drive.
RAID 1: Mirroring Without Parity or Striping
In a RAID 1, two or more disks are configured so that data written to one disk is copied exactly to the others. This configuration doesn’t increase storage capacity, but it does increase read speeds and keep data secure, should one of the drives fail. The maximum size of a RAID 1 is equal to the total capacity of the smallest drive on the array. For this reason, it’s usually advised to keep the drive sizes consistent, but not required. Because of the boosted reliability provided by mirrored data sets, the more drives you add to a RAID 1, the more reliable it will become. If one disk has a failure rate of 5% over three years, a paired set in RAID 1 will have a failure rate of 0.25%.
RAID 5: Striping with Distributed Parity
RAID 5 is one of the most popular levels of RAID used today. In RAID 5, data is distributed evenly at the block level between three or more drives. A parity of each block of information is also placed on a separate drive than the original block, allowing one drive in the array to fail without risking data destruction. During the period where a RAID 5 is running with a failed drive, it becomes as reliable as RAID 0, requiring a replacement drive in order to rebuild data structure. RAID 5 gained mass appeal due to its ability to get more storage capacity out of the combined drives, allowing you to get 3 TB out of 4 x 1 TB drives rather than 1 TB or 2 TB possible in RAID 1 and RAID 1+0.
RAID 6: Block Striping with Double Distributed Parity
RAID 6 works very much like RAID 5, with an extra parity distributed between drives in the array. This means that you can experience two drive failures at the same time and still have access to all of your data. Because of the extra parity, the write speed to the array takes a minor speed decrease, and the minimum number of drives in the array increased from three in RAID 5 to four. The available storage capacity of a RAID 6 with 4 x 1 TB drives is 2 TB.
RAID 10: Hybrid Combination of RAID 1 and RAID 0
RAID 10, also known as RAID 1+0 is a combination of RAID 1 and RAID 0 configurations that requires a minimum of four disks (increasable in pairs) to set up. In a RAID 10 with four drives, each pair of drives act as RAID 1 arrays, giving mirrored redundancy to each block of information being written between them. Each pair is combined under a RAID 0 configuration, receiving every other block of information as it is written. This, in theory, provides the speed boost found with RAID 0 and the fault tolerance of RAID 1. In a RAID 10, two drives can fail without loss of data, as long as they don’t exist within the same mirrored pair. It is currently widely used in database, email, and Web server capacities due to the extremely high performance and tolerable failure risk levels.
RAID 50: Hybrid Combination of RAID 5 and RAID 0
With at least six drives required to operate, a RAID 50 provides a wide range of possible configurations that vary greatly in terms of storage capacity and failure tolerance. In RAID 50, multiple sets of RAID 5 sets exist under a single RAID 0 body. Data written to this RAID is distributed between multiple RAID 5 sets, with each maintaining a level of parity. Fault tolerance with RAID 50 varies depending on how many RAID 5 sets exist within the array. If you have nine RAID 5 sets, you can theoretically lose nine drives simultaneously and still maintain complete access to all of your data, but the failed drives must be replaced in order to rebuild the data structure and return to optimal performance and reliability levels. The more drives you add to RAID 50, the more capacity and overall reliability exist in the array, at the cost of an increased fault recovery time.
There are quite a few more RAID configurations worth noting, and there is a significant amount of debate going on in the IT industry over which array is best for various purposes. Creating a balance between performance, storage capacity, and reliability is a matter of determining what the best tool is for the specific task at hand.