Page 66 - UNI 101 Computer Science Handout.

Faculty of Nursing
Adult care Nursing Department

In a striped array, a file segment to be stored is divided into blocks. Different blocks are then written

simultaneously to different disks. This effectively multiplies the throughput rate by the number of data

disks in the array. A striped array requires a minimum of three disk drives; in the simplest configuration,
one disk drive is reserved for error checking. As the write operation is taking place, the system creates a

block of parity words from each group of data blocks and stores that on the reserved disk. During read

operations, the parity data is used to check the original data. There are five well-defined RAID standards,
labeled RAID 1 through RAID 5, and a number of additional proprietary and nonstandard varieties,

including one labeled RAID 0.

The most common of these are RAID 0, RAID 1, and RAID 5. RAID 1 is a mirrored array as described

above. RAID 1 provides protection by storing everything at least twice, but offers a substantial
performance gain, particularly under heavy data read usage. RAIDs 2, 3, and 4 are arrays that are striped

in different ways. Each uses a separate disk for error checking. Since data on every disk must be checked,

this can create a roadblock on the single disk that is used for error checking. RAID 5 eases the roadblock

by spreading the error-checking blocks over all of the disks. RAID 0 is not a true RAID, because it provides

no redundancy and no inherent error checking. Data is striped across all of the disks, primarily for fast
access. However, the lack of redundancy means that a failure of any single disk block in the array

corrupts all of the data in the system. However, this shortcoming can be overcome with proper backup

and with certain types of journaling file systems. It is also possible to ‘‘nest’’ RAIDs. For example, we can
use a pair of RAID 0 groups inside RAID 1 to achieve mirrored redundancy. The combination is known as

RAID 0+1. With or without the additional protection, RAID 0 is sometimes attractive as a low-cost

method of achieving high data transfer rates when they are required. A number of vendors provide RAID

controller hardware. Particularly for largeRAID5 systems. With RAID controller hardware, RAID

processing takes place within the array controller. The array appears as a single large disk drive to the
computer. It is also possible to create a RAID using conventional, off-the-shelf disk controllers and

operating system software. Although this uses CPU processing time, modern computers have enough

spare power to make this a practical solution in many instances. It also reduces the possibility that a
single RAID controller can cause the entire array to fail.
66 Academic Year 2025/2026