RAID Explained


RAID is not backup, nor does it replace a backup strategy. Backing up to a RAID device might well be a part of such a strategy. RAID can be a great way to optimize NAS and server performance and quickly recover from hardware failure, but it’s only part of an overall disaster-recovery solution.


RAID stands for Redundant Array of Inexpensive (or “Independent”) Disks. A RAID-enabled system uses two or more disks to improve fault tolerance for a machine (computer). Fault tolerance simply is ensuring a machine with a failed component (usually a hard drive), can still operate. The advantages of fault tolerance are less interruptions in productivity,  decreased the chance of data loss, and ensuring your buisness does not have downtime.


Here we will explain the different types of Raid terminologies and what they mean.

This is the simplest kind of RAID.  In this system Data is written to multiple disks.  Think of the phone company you could have one disk for Washington, Another for New York, Anotther for Los Angeles.  Read and write operations are extremely fast.  The downside is a failure of one disk will bring the whole system down.  We hardly see this setup being used for critical systems anymore as the cost savings is less than $100.00.

This is typically the set up we use for home businesses/small business.  The idea is the system copies all the data written to another similar disk in the system.  If one drive fails you will be alerted but you can continue with your work as the system will simply switch to the working harddrive.  The disadvantage of this is speeds are slightly slower than a RAID 0 (to be honest we are talking about 5 seconds a day here) system and you only have 1/2 the storage space available to you.

Raid 5 is typically found in medium enterprise solutions. The idea here is 3 disks up to a maximum of 16. How this works is computers work on binary numbers. The controller takes the two blocks of data and performs a calculation on them to get a parity value denoted by the P block. This is based off raid 4 where all the parity values are written to one disk. The advantage here is that you only lose 1/3 of the space to the backup data. If one disk fails it can be replaced. However large data sets can take a month to replace and its easier to just shut the computer off and do a restore. In theory we can just replace the failed drive and have the controller rebuild it through algebra but the performance will be significantly affected. And the replacement has to be immediate otherwise the system will be down.

Most people call this setup Raid Ten. However, the correct way is Raid One and Zero. This is the setup usually seen in very large organizations. You lose 1/2 the space to the backup. However you get increased performance and any one drive failing does not have catastrophic consequences. Theses systems are usually hot swappable meaning when the drive is identified you simply pull it out and replace it with a new one and life goes on. The controller will just replicate the new drive with the still functioning drives data (1 hour) with no noticeable performance issues.