RAID is constituted by four mysterious letters. We use them every day when we talk about NAS but few people know what it is exactly. The meaning slightly changed during times, and now it stands for Redundant Array of Independent Disks. What does that mean? Let’s have a look to the dictionary.
Redundant: Not strictly necessary to functioning but included in case of failure in another component.
Array: An ordered arrangement, in particular
Independent: Not connected with another or with each other; separate.
Disks: An information storage device for a computer.
If we sum-up everything, a RAID is an arrangement of separate information storage device, or hard disks, which are not strictly necessary to functioning but included in case of failure in another hard disk.
Flexibility – the RAID levels
There are a lot of different RAID levels which are designed to specific situations: do you need more space or more safety? Do you need to have more power? How many hard disks can be lost before data loss? Here is the list of the most common RAID levels.
JBOD: that’s not even RAID!
JBOD stands for Just a Bunch Of Drives. It’s pretty clear: all your disks in your NAS are totally independent and no one is helping the other. This configuration is really not useful for professional usage or only in very specific cases.
This RAID is very specific: in its native form it boosts performances by writing data on every disks at the same time. This way, no time is lost during this process and writing times are getting higher and higher with the numbers of disks. However there is a major catch: you lose only one disk in the whole array and all your data are lost! This raid level is now used for nested levels; we’ll arrive there a bit later.
RAID 1: slow but with a growing reliability/inefficiency
Also called mirroring RAID, this one writes the same data on all the disks in the array as a mirror. With a lot of disks, writing can be a by long and there is not reading benefits. This level is very convenient for small arrays (2 disks) or for very sensitive data: as long as one disk is safe, all your data are safe! Due to a very low efficiency, it’s not commonly used in big arrays.
RAID 5: the compromise between safety and performances
This one is a bit more complicated but is also one of the most popular RAID configurations. It requires at least 3 disks in the array: that’s why Thecus was the first to bring RAID5 to home usage with 3 bays NAS.
The RAID5 is writing what we called a “parity block” every n blocks, with n-1 the amount of disks in the array. For a 3 bays NAS, it would be every 2 blocks. Each block is written on only one disk. This way the writing speed is quicker than RAID1. But what happens if one disk fails? Easy: the RAID controller will rewrite every block on the new disk. If it’s a parity block, no data is lost and if it’s a data block, it can be calculated using the parity block on other disks. This process is called rebuilding the RAID.
However, this solution got drawbacks: only one disk can fail at the same time (two failures at the same moment happens more than people think!), the rebuild time can be a big long even if the NAS is still available during this time, and it can get really slow for big arrays. But the very high space/safety efficiency made this RAID level one of the most popular among NAS users.
RAID6: safer than RAID5, but a bit slower
It’s an option to consider since you should need less a spare drive. Performances are also not as good as a RAID5 array.
Nested levels: RAID10, 50, 60…
Those RAID levels are a bit more complicated. We should actually say 1+0 for 10, 5+0 for 50… Most of the time it safe and slower arrays in RAID 1 or 5 which are themselves used as a disk in a array in RAID0.
|RAID Level||Actual Space||Fault Tolerance||Reading||Writing|
|RAID0||n disks||0 disk||Good||Good|
|RAID1||1/n disks||n-1 disk(s)||Good||Bad|
|RAID5||(1-1/n) disks||1 disk||Good||Medium|
|RAID6||(1-2/n) disks||2 disks||Good||Medium/bad|
Example of the available space with some disks configurations