In the last installment I explained my intention of install and configure a RAID from 0 to finally dispose of a NAS, so in this installment we will see which are the RAID types and categories there are.
It's a RAID?
A Redundant TOrray of Independent Disks or a redundant array of disk independent is a storage system consisting of two or more physical disks to our operating system sees as if they were a single logical unit.
In this article I will explain the types of RAID that today you are being implemented both at the enterprise level as domestic, Although in Wikipedia you have the description of all and each one of them.
So they serve the RAID?
Depends on the type of RAID We choose we can serve for one thing or another. Some RAID keep data parity, so in case that one of the disks is damaged, by adding one new and blank the system, We can restore it and not lose any data, other arrays do not hold parity, but to read and write simultaneously on multiple disks that the operating system sees as one single, We gain speed of reading and writing in the file system.
Other types of RAID combined the two above options.
Normally the disks or partitions that make up a RAID must be of the same size.
Nomenclature of logical drives RAID-level system in GNU/Linux
In GNU/Linux logical units corresponding to arrangements with various software-based disk follow a nomenclature /dev/md * instead of /dev/sd * that standard is used for partitions.
Categories of RAID
The RAID they can be configured by Software or by Hardware, is a program that runs from above the SSystem Operativo, or the firmware of a RAID controller hardware that works independently to the SO.
It is also possible to buy in a store a NAS/RAID with capacity for two or three discs with everything configured, but as explained above, I prefer to do it myself and gain in safety and modularity.
The RAID by Software It is that you can configure through a programme, i.e., a Software package that runs on top of the operating system, like for example mdadm on GNU/Linux or the New Volume Wizard RAID in Windows.
An of the advantages This category of RAID is that if the operating system fails, We can connect storage devices, hard drives in my case, to any other PC with the same base operating system, and have the information immediately.
The RAID hardware is that is configured by using a physical device that acts as a controller of an independently operating system disks.
Is normally configured through of a Firmware can access any of the keyboard function keys, normally F11 o F12, at the start of the PC and from there set up the RAID so that the operating system recognizes it directly as a single physical disk.
Some motherboards (above all the most modern) they have support for RAID in some or all of your connectors SATA.
The specifications indicated the types of RAID that supports, and one of its advantages It is the increase in the speed of read/write, but If the controller card is damaged or any of its ports, We will have to find one with the same Chipset to access the information.
The RAID 0 or striped volume equitably distributes the file system data between two or more disks, This type of RAID It is not redundant, so if failure one of the albums that compose it will lose data.
Its intended use more common is to increase the performance of the system, or you can also reuse of low-capacity volumes creating with them one of greater size and performance, Since we wrote and read simultaneously in all disks at the same time.
With this type of RAID the logical volume is the size of the sum of the disks that make up the, for example, if we create a RAID 0 with two floppy-disk of 320GB will have a logical device from 640 GB.
The RAID 1 or mirror volume It maintains an exact copy of the same data across all physical disks that comprise it, that means that this RAID If it is redundant, so we will only lose data in the strange case that damage the entire physical disks that make up the.
Its purpose more common is the to try to ensure the integrity of the data, increasing the rate of initial reading and writing, in the worst case by keeping an average of the rates of physical disks that make up the.
In this type of RAID We could say that in Exchange for reliability, we lose the possibility to take advantage of the size of the physical drives that compose it, Since in this case when there is redundancy, If the system we will create two albums, or two partitions of our 320GB RAID will be also 320 GB.
With the advantage of if one of the disks is damaged, only have to replace it and pair it with the other to thus restore the RAID without losing any data.
The RAID 5 or distributed parity RAID distributes the data in blocks hosted on different physical disks, but each disk stores a parity of any other disks block, which is how a “Summary of information containing another disk”, or something like a checksum.
This type of RAID requires at least 3 physical disks to be implemented, and it has no theoretical limit of units, but is usually limited to prevent data loss, in the next paragraph we will see why. The more it is normal to implement it through RAID hardware, and does not provide too much performance, because every time you write in any disks, is recalculara block parity corresponding to that block we have modified.
Parity blocks are only read when there is an error in any sector with data and it should be restored, also if more than one disk fails at the same time we will lose all of the data since it will be impossible to restore its contents by the lack of parity blocks.
The RAID nested are made up of others RAID instead of physical disks and allow us to combine the advantages of different types of RAID.
We can represent them as layers, with the RAID nested in the upper layer, the RAID in the intermediate layer, and the physical disks in the lower, and they tend to combine types of RAID redundant to ensure the integrity of the data, with RAID 0 to increase performance.
Due to the large size of the current disk, the probability of failure in these RAID types is growing.
The RAID 0+1 or mirror divisions, used to replicate and share data between different disks.
To implement this type of RAID, First we must create two sets of RAID 0, to finally implement a RAID 1 with them, so when a physical volume fails, the lost data can be recovered from the other set RAID 0 to rebuild the global set.
But if you want to add a disk but the whole, We will have to do so in each of the subsets, so disks must be added in pairs in case of large, Furthermore this type of RAID only supports the failure of one of their albums at the same time, is to say that if each array disk fails RAID 0, or two of the same array, We will lose all the data and if one of the drives fails all the discs in the set they will participate in the reconstruction of the whole.
The RAID 10 or RAID 1+0, also known as division of mirrors is similar to a RAID 0+1 except that in this case we are implementing an RAID 0 from two RAID 1.
In each division RAID 1 one of the two disks can fail without losing data, but the failed disk should be replaced immediately, as if a single disk for one of the RAID 1, the RAID 10 passes to act as a RAID 0 which means that we will lose the entire data if a disk fails.
To expand this type of RAID also we have to add two disks.
If you like you can follow me on Twitter, Facebook, Google +, LinkedIn, or share it with the buttons under this publication, If you have any questions or suggestions please do not hesitate to comment.