A Short Introduction to RAID Technology

RAID (Redundant Array of Independent Disks) is one of many fault tolerance methods available to information technology specialists. It can be used in large file servers or on desktop PC systems to provide data redundancy and performance enhancement. It is available through a wide variety of vendors, including: 1) Seagate, 2) Maxtor, 3) Western Digital, and 4) Hewlett Packard.

RAID Defined

RAID is an array configuration using multiple hard disks to provide fault tolerance and improved access rates. Hard disks configured with RAID save and read from an array of hard disks as if they were one disk. This is accomplished by reading and writing in stripes, overlapping I/O operations in a balanced way to improve performance. Raid offers redundancy by storing the same data in different places on multiple hard disks.

When configured into a RAID array, the operating system treats the array of hard disks as only one hard disk, partitioning the space on each drive into units that measure from 512 bytes to several megabytes. By setting the size of the stripes so that a single record spans all the disks in the array, performance is improved, since the system can access and read all the disks at once. Regardless whether it is a single- or multi-user system, the best performance is attained by configuring the stripe wide enough to hold the maximum size record.

Levels of RAID

There are many single levels of RAID, as well as combinations of the levels, including:

1. RAID 0: The controller or program stripes data across multiple drives of the system to provide improved performance.

2. RAID 1: The controller or program duplicates all data saved to one drive to another, mirroring all data to give data redundancy. If one drive fails, the data is duplicated on the other.

3. RAID 3: Level three of RAID stripes the disks and provides data redundancy through a dedicated parity drive. This level requires three hard disks, two for striping and one to be the dedicated parity drive. All data is managed at the block level, making RAID 3 ideal to handle large data files and image files.

4. RAID 5: RAID 5 uses block level striping and distributed parity. Parity blocks use blocks of data written across the drive array. By separating the parity information for a data block form the data on the other drives, lost data can be rebuilt from the data on the other drives. This level, an excellent combination of performance, redundancy and storage efficiency, is one of the most popular RAID levels.

5. Combinations: When one level does not provide the needed functionality, levels can be combined to address the need for increased performance. Most of the time, this must be done utilizing a hardware RAID controller. Software solutions are seldom capable of handling the increased level of complexity of the combined RAID levels.

Implementations

There are two forms of implementation for RAID, hardware and software. Hardware RAID controllers interface with the system and hard drives through SCSI or IDE/ATA. The hardware solutions can be either internal or external controllers. Software RAID is most often used on desktop PC systems with multiple drives.

Data Redundancy and System Performance

RAID is an excellent technology to provide fault tolerance to today’s systems. It is not only provides redundancy of data, it will increase system performance. Any company with business critical data storage needs should investigate, and invest in, RAID technology.