RAID or Redundant Array of Storage Discs stores the same data over different places. Multiple hard discs store this data and protect it in case of a drive failure. Redundancy is, however, not present at each RAID level.
RAID is characteristically used to upkeep the speed and performance of the disc storage of a computer. It is very frequently used over high-performance computers and servers.
History of RAID
The term RAID first came to prominence in 1987. It was penned down in a 1988 technical report by David Patterson, Randy Katz, and Garth A. Gibson. The report was titled “A Case for Redundant Arrays of Inexpensive Disks (RAID).”
The authors expressed the opinion that the performance of top disc drives of the time could be improvised upon by using an array of inexpensive drives. With the use of redundancy, the reliability of a RAID array supersedes the reliability of any single disc drive.
The term ’inexpensive’ in the acronym was replaced with the term ‘independent’ by industry vendors. This was to overlook the implications of being low-cost.
How does RAID work
By using RAID, data is placed over multiple discs. Correspondingly, the input/output operations overlap in a balanced manner. This, in turn, improves system performance.
To an operating system, RAID arrays seem like a single logical hard disk. The technologies used for RAID’s operation include disk striping and disk mirroring.
With mirroring, two RAID arrays can have two drives that contain the same data. So, one drive continues to work even while the other crashes.
Similarly, striping partitions a drive’s storage space into units. A unit may be a sector or 512 bytes. It may alternately be several megabytes of storage space. This process is striping.
Stripes of each of the discs are interleaved. They are addressed in the defined order.
Sizes of the stripes depend on the application for which RAID is used. For a single user system that involves large records, the stripes are characteristically smaller. This ensures that the record spans throughout the discs. It also makes the record easy to access, and all discs can be read at the same time. An example of RAID wherein smaller stripes are used is for storing scientific images or medical records.
Stripe size, however, must be wider in the case of a multiuser system. This allows the strip to easily accommodate a record of the maximum size. Correspondingly a user can avail overlapped disk I/O across drives.
Disc striping and disk mirroring can both be combined over a RAID array. They are both used together in RAID 01 and RAID 10.
A RAID Controller defines the level of abstraction between physical discs and the OS. So it protects data in the event of a crash and improves performance as well.
A RAID controller can be used for both, hardware and software-based RAID arrays. For a hardware-based RAID product, the array is managed by a physical controller. This may be in the form of a PCI Card, or it may be the part of the motherboard.
If a RAID array is software-based, the RAID controller has the same functionality as a hardware-based RAID array. But it uses the resources of the hardware system.
Benefits of RAID
The prime benefits of RAID are cost-savings, resiliency, and performance. RAID is characteristically an improvement over the operation of a single hard drive. It is subject to how RAID is configured.
Following a crash, RAID enhances computer reliability and speed.
With RAID 0, files are split up. They are then distributed across drives. The drives all work upon the same files. Read and write operations are correspondingly quicker as compared to using the same drive.
With RAID 5 as well, array data is broken into sections. But another drive is dedicated to parity as well. When one non-parity drive fails, the parity drive sees what is working. The parity drive hence figures out the data that was present over the failed drive. As a result, RAID 5 features a higher availability.