RAID

Introduction

RAID (Redundant Array of Innexpensive Disks) provides computer users with the ability to configure a set of hard drives to work together as one. Depending on the RAID level in use, these arrays can increase performance, protect data or both. This page contains a detailed summary of the various RAID levels currently available to end users.
[edit]
RAID Levels

Each RAID level operates in a different manner in order to address the needs of the many different groups of users who need to use drive arrays. These levels are detailed in the following section in order to allow users to figure out which one (if any) is suitable for their requirements. Combined, these different levels cover a wide range of scenarios that users requiring drive arrays will find themselves in.
[edit]
Level Summaries

[edit]
RAID 0

Interleaves data between a number of drives in order to maximize the performance of the array. While this level provides the highest possible performance, it has no redundancy and actually increases the risk of data loss. Generally used for scenarios where performance is crtical and the data can easilly be reconstructed. See the dedicated article for more information.
[edit]
RAID 1

Stores an identical copy of all data on all of the drives contained in the array. This level provides the highest level of protection, as all but one of the drives in the array can fail without any loss of data. Unfortunately, it provides no performance benefits and only provides the capacity of a single drive in the array (ie two 160GB drives in a RAID 1 array will only provide 160GB of capacity). See the dedicated article for more information.
[edit]
RAID 3

Operates in a manner similar to RAID 0, but adds a set of parity information saved on a seperate drive. This provides performance levels similar to RAID 0, but allows a single drive to fail without losing any data. Unlike RAID 0,4,5 and 6, RAID 3 interleaves data on a byte-by-byte basis. See the dedicated article for more information.
[edit]
RAID 4

Works in the same manner as RAID 3, but interleaves the data using larger blocks of data. See the dedicated article for more information.
[edit]
RAID 5

Works in a similar manner to RAID 4, but distributes the parity information across all of the drives rather than using a dedicated drive. This generally provides better performance when the array is operating with a failed drive. See the dedicated article for more information.
[edit]
RAID 6

Works in the same fashion as RAID 5, however it stores two sets of parity information so that the array can survive the failure of two hard drives without losing any data. This is a more secure form of data security, but can significantly increase processor overhead. See the dedicated article for more information.
[edit]
RAID 10, RAID 0+1

Nested RAID modes that combine RAID 1 and RAID 0 to provide both redundancy and performance advantages. See the RAID 10 and RAID 0+1 articles for more information.

[edit]
Comparison Table

Level	Performance (4 Drives)¹	Survivable Failures²	Efficiency (4 Drives)¹	CPU Overhead	Minimum Drives
RAID 0	400% (N)	None	100%	None	2
RAID 1	100% (1)	All but one	50% (1/N)	None	2
RAID 3	300% (N-1)	1	66% (N-1/N)	High	3
RAID 4	300% (N-1)	1	66% (N-1/N)	High	3
RAID 5	300% (N-1)	1	66% (N-1/N)	High	3
RAID 6	200% (N-2)	2	66% (N-2/N)	Highest	4
RAID 10	200% (N/2)	1+³	50% (N/2)	None	4
RAID 0+1	200% (N/2)	1+³	50% (N/2)	None	4

¹ All comparisons are done with a four drive array to provide an even basis for all levels. As these figures may be different for arrays of different sizes, a formula has been provided to calculate performance in an N drive array.
² The number of drives in the array that can fail before data will be lost.
³ Depends on the configuration of the array. See the RAID 10 article for more information.
[edit]
RAID is Not a Backup

While some RAID modes provide a high degree of redundancy, it is not a replacement for proper backup procedures. Redundant RAID modes only protect users against failures in the physical hardware. If, for instance, a machine is infected with a virus that destroys the files, that damage will be immediately reflected on all copies of the data stored in the array.
Similarly, any catastrophic events like fires or floods are likely to damage all components inside of a computer equally. All RAID modes require at least one drive to remain intact in order to retreive data, so loss of the entire array means loss of all data stored on it. Conventional backups stored in another location may survive these incidents and allow users to retreive their information.
As such, regardless of the RAID configuration that is in use it is important for users to maintain a proper backup strategy. The primary benefit to RAID arrays is that they can continue to opperate after a drive failure - this allows users to address the problem at a time of their choice rather than being forced to deal with it right away.
[edit]
Software vs. Hardware RAID

With the popularization of RAID, there is now a wide range of controllers available on the market these days. While they all perform the same basic task, the way that they work under the hood can be very different. While there are many factors at play, the most significant of them is whether they are hardware or software based.
[edit]
Hardware RAID

Traditionally, RAID controllers have primarilly been driven by hardware designed specifically to deal with the computational tasks required to maintain their arrays. These hardware-based controllers often had dedicated RISC microprocessors to handle computationally-intensive tasks such as generating parity information for RAID 3,4,5 and 6 arrays.
At this juncture, hardware-based controllers are generally isolated to the high-end of the market. The additional electronics needed to handle these tasks can add significant complexity and, in turn, increases their price. Whether that added cost is justified depends on how the array will be used and which RAID mode is in use.
[edit]
Software RAID

As vendors have pushed RAID controllers into the mainstream, the easiest way to cut costs was to offload some tasks to the computer's main processor. By doing this, the complexity of the controller can be significantly decreased allowing the cost to fall to commodity levels. This, in turn, made it practical to include basic RAID functionality in mainstream motherboards.
In many single user scenarios, this will often not have major negative effects on the array's performance. Simple RAID modes are relatively easy to implement, so the overhead on the main processor is generally relatively small. In scenarios where a large number of users will be simultaneously accessing data stored on the array, or where more complicated RAID modes are used (eg RAID 5) this load can become significant.
[edit]
Grey Areas

As with all things, not everything is black and white. There are many controllers that use both hardware and software concepts to manage the array. For instance, many mid-range controllers handle many tasks in hardware but leave the more complex tasks (eg RAID 5 parity generation) to the processor. The end result is that these controllers will provide performance between the two extremes, allowing users more granularity in selecting their best balance.
[edit]
See Also

RAID 0 - Stripes the data at the block level to maximize performance, however that increases the risk of lost data versus individual disks.
RAID 1 - Mirrors all data onto all of the drives contained in the array. This provides the highest level of protection, however it is also relatively inefficient.
RAID 2 - Similar to RAID 0 but stripes at the bit level rather than the block level. This level is very uncommon and not supported by any modern RAID controllers.
RAID 3 - Uses byte-level striping with a dedicated parity disc. This can provide similar performance to RAID 0 but can survive the failure of one drive.
RAID 4 - Same as RAID 3, but uses block-level striping.
RAID 5 - Uses block-level striping with a distributed parity disc. This can provide similar performance to RAID 0 but can survive the failure of one drive.
RAID 6 - Uses block-level striping with two distributed parity discs. This can provide similar performance to RAID 0 but can survive the failure of up to two drives.
RAID 10 - A combonation of RAID 0 and RAID 1. Data is striped across two RAID 1 arrays. This provides the performance advantages of RAID 0 and can survive at least one drive failure.
RAID 0+1 - A combonation of RAID 0 and RAID 1. In this mode, two RAID 0 arrays are mirrored. This provides the performance advantages of RAID 0 and can survive at least one drive failure.
Matrix RAID - A propreitary technology used by some Intel chipsets that allows drives to be partitioned and different RAID levels used on each block.