RAID – Redundant Array of Independent Drives or Redundant Array of Inexpensive Drives.
A RAID is a method of storing data that uses more than one disk to appear as a single disk. There are several different types of RAID which utilize multiple disks in different ways. By arranging the disks in different ways different RAIDs have different benefits, including:
- Increasing the size of storage
- Increasing speed of the storage media
- Increasing the resilience/reliability of the storage
How are RAIDs connected to a computer?
There are a variety of different ways of building a RAID and attaching it to a computer, but they fall into two main categories– Software RAIDs and Hardware RAIDs.
Hardware RAID: A hardware RAID is a dedicated unit that the hard drives are placed into. This unit then attaches to a computer (normally a server) via one of several connections (normally SCSI or Fiber). The RAID unit then “presents” the RAID unit to the computer. For example, the RAID unit may have 8 HDD drives in, e.g 500GB in size, but the computer would only see 1* 4 TB (Terra byte) drive, as the RAID has combined 8*500GB drives to form 1 4000GB/4TB drive. The computer does not know, or care that there are only 8 drives and will see 1 logical unit presented by the RAID.
A software RAID is where a host computer does all the work of converting multiple drives into a RAID. For example a user can place multiple drives into Windows XP Pro computer, and then request that Windows combine them together – using the Dynamic Disk function. The net result is that the user is presented with “volume” that can consist of multiple drives.
To the end user the Software and Hardware RAID may appear the same, i.e multiple drives are taken and presented as a single drive. The difference is cost and reliability. A dedicated RAID unit, without any drives, can easily costs over £2,000, just for a 8 drive bay, with a SCSI connector (without any drives). This is a dedicated unit that will handle all the requests in relation to the RAID, it can manage the volumes, record activities on the RAID, and alert the administrator to the errors. A software RAID, such as the one mentioned on windows will run on any hardware, as long as Windows XP is installed, it is simply not in the same level of performance. For home use, a software RAID is perfectly fine. For business a dedicated RAID is normally required for speed and reliability.
Types of RAID
The drives in a RAID can be arrange in several different manners, however the three main ways in which they are used are as follows:
RAID 0 – This is also known as a “stripe” and takes a minimum of two disks to work. This spreads the data across the drives, e.g 2*250 GB hard drives would appear as 1*500 GB hard drive. If multiple reads and writes are occurring at the same time – e.g two people accessing different hard files, then a strip will be faster to access than a single hard drive. This is because the data can be accessed from multiple locations at the same time. i.e there are two heads reading at the same time rather than one. However, there is no “redundancy” in the a RAID 0. Therefore is one drive fails all of the data could be lost.
RAID 1 – This is also known as a “mirror” and takes a minimum of two disks to work. This is the simplest of RAIDs to understand, for every disk in the RAID another disk is an exact mirror, e.g if there are 2*500 GB hard drives, the RAID unit presents just 1*250 GB hard drive. The other is not “seen” by the user, but it constantly replicates the primary disk.This means that if one hard drive fails, the other one will take over, it is quite possible that an end user would not be aware that a hard drive had failed and they could continue working as normal. While a mirror is highly resilient, is is inefficient in its use of media, as there is a 100% redundancy.
RAID 5 – This requires a minimum of 3 disks. This has the benefits of both RAID 0 and RAID 1, and less of the disadvantages. In a RAID 5 the data is spread across the disks, have one drive acts as “spare”. This means there is an increased size, and increased speed, and increased redundancy.
For example, a RAID unit with an 8*500 GB disks in a RAID 5 array would present a single 3500 GB (3.5TB) volume to the host computer (7*500). One of the 500 GB drive drives does not count to the volume size as it provides redundancy. However, and this is were the RAID 5 is very clever, all of the spare data is not stored on a single hard drive (as it could not do that), rather it is spread across all of the other drives, evenly, taking up a total of 1 drive out of the unit. This means that if any one drive on the RAID 5 fails, the system can continue without any noticeable effect to the user (though, depending on the RAID access times may decrease). The RAID 5 achieves this by using something called “parity”, this is a more detailed topic and will be covered later this month.