The Nature of the Beast

If you have a single hard drive, that would suffice for temporary purposes. However, you will, eventually, lose that data because that hard drive will fail. Whether it is Hard Disk based, or Solid State, your drive will fail. Either the Hard Disk Drive will fail because of mechanical components breaking or wearing out, or that Solid State Drive will stop working altogether.

                When that Hard Drive fails, and it will, all data on it is to be considered lost. There are Data Recovery Specialists that, for a large fee, can get that data back. However, these services depend on HOW the drive failed. For example, if the head of the Hard Disk Drive gets stuck on the disk (which can spin 5,400, 7,200, or 15,000 times per minute), the data on the drive is gone and no one can extract it. But this isn’t the only way drives can fail. Also, there is a solution to this inevitable failure of all drives.

                The solution to the inescapable failure of all drives is to have more than one drive, or a Redundant Array of Independent (or Inexpensive, depending on region) Disks (R.A.I.D. for short). And there isn’t just one type of R.A.I.D., There are (arguably) 8 types of Standard R.A.I.D., with 4 of those types being most common. Most types of RAID require at least two Identical drives to function.

JBOD

JBOD, or Just a Bunch of Disks, also known as Spanning, is a type of raid where data is written to a drive until it is full before writing data to another drive in the array. JBOD allows for no losses in data storage volume, but has no redundancy, and no increase in read/write speeds. Because of the lack of redundancy, JBOD is I considered a type of RAID, and should not be used in storing vital data.

RAID 0

                RAID 0, or Striping, is a type of RAID where block one of a file is written on disk one, and block two of that same file is written on disk two. This is a way of increasing the speed at which a file is written to (or read from) the drive, however, it does come with a cost: redundancy. If you lose any drive in the array (of two or more drives) all data is lost. Because of the complete lack of redundancy, this RAID is not considered a good solution for important data.

RAID 1

RAID 1, or Mirroring, is a type of RAID where data is written as exact duplicates to two (or more) drives simultaneously. If you have a good drive in the array, your data is safe. RAID 1 does, however, take a performance drop when writing, because each drive must be written to simultaneously. However, when reading data from the drives, you will see a read rate increase, similar to that of RAID 0.

RAID 2

RAID 2, is rarely used, and stripes differently than any other RAID type. In RAID 2, The drives must all spin at the exact same speed and must all start at the same time. This ensures that the drives reach the same point on the disks at the exact same time. Also, RAID 2 can only service one request at a time (meaning, it can read a file, or it can write a file, but cannot do both at the same time). RAID 2 also requires use of “Hamming Code” for error correction. Extreme data transfer rates are possible.

RAID 3

                RAID 3, also rarely used, is very similar to RAID 2, in that data is striped across multiple drives. However, there is also a Parity Block that is stored on a dedicated drive and used only when a drive fails. Parity Blocks are the sum of a block of data and will tell the device WHAT data is missing in the event of a drive failure. RAID 3 was replaced with RAID 5.

RAID 4

                RAID 4, also replaced by RAID 5, was block level striping, and a dedicated Parity Drive, which, like RAID 3, stored a sum of the blocks on other disks, used when a drive fails.

RAID 5

                RAID 5, or Striping with Parity, stripes files across multiple disks, and uses a parity block. However, this parity block could be in any drive, and is typically shared between all drives. RAID 5 does, however, require the use of at least 3 disks. When writing to the drive, it is, in theory, just as fast as RAID 0, because the data and Parity is Striped across multiple drives. This allows for one drive to be lost and can function (although slower) with a missing drive. The only loss is volume: the parity blocks take up one whole drive worth of storage space.

RAID 6

                RAID 6, or Striping with Double Parity, is like RAID 5, in that data is striped across all drives. There is, however, a second parity block stored. With RAID 6, one can lose a total of 2 drives in the array before any data is lost. The only downside is lost volume: you lose 2 drives worth of space in parity blocks.

Nested RAID

                Nested RAID, or Hybrid RAID, also comes in various types. The only similarity between them is that they are two types of RAID smashed together. RAID 01, is Mirrored Stripes (two RAID 0’s smashed together to create redundancy). RAID 03 is three RAID 0’s, one of which being Parity. RAID 10 is a Stripe of Mirrors, where two RAID 1’s are mirrored to create redundancy. RAID 50 is three RAID 5’s, striped, with one RAID 5 devoted to parity. RAID 60 is like RAID 50 and requires 8 drives. RAID 100 is two RAID 10 arrays (two RAID 1 arrays each), in RAID 0 Configuration.

Nonstandard RAID

                There is a myriad of RAID Types from various vendors, the most common of which are RAID Z, and SHR/SHR-2. RAID Z utilizes a specific file system and requires no special hardware. SHR and SHR-2 will only run under Synology Disk Station Manager® Operating system.

Sources:

Memory Jogging Source: https://en.wikipedia.org/wiki/Standard_RAID_levels

Glanced at this Source: https://en.wikipedia.org/wiki/Non-standard_RAID_levels

Barely Looked at this Source: https://en.wikipedia.org/wiki/Nested_RAID_levels