i often get private messeges with questions about RAID. Most of them along similar lines.
I thought I'd take the time to go over some of the fundamentals of RAID. This will prove useful to people who are new or have a basic understanding of RAID. Please read...
RAID stands for Redundant Array of Inexpensive Discs. It is a technology which has been used for SCSI hard drives for some time now. Most computers use IDE drives, and many are now turning to using IDE raid.
What RAID is used for...
RAID is a technology with two main purposes. Firstly, it enables performance enhancements, secondly, it helps prevent data loss. There is no one standard for RAID. There are different standardised 'levels' of RAID. The most common are known as RAID 0 and RAID 1. For simplicities sake, I will explain only these.
RAID level 0 is for the soul purpose of improving performance. RAID 0 works by using multiple hard drive as a single entity. When a file is written to disk, parts of the file are written to each disk in the RAID array. If you have two drives in a RAID array, they would simultaneously write half the data, theoretically halving the time it takes. This applies to reads also. A RAID array can be configures to chunk sizes. An 8kb chunk size alternates 8kb of data to each disk in the RAID array.
The problem with RAID 0 is that there is no fault tolerance. If just one drive in the array fails, you have lost all your data. The benefits of RAID 0 are that you get all the space you pay for. EG, 2 40Gb drives in RAID 0 total 80Gb of space.
It is advisable to use to use the same drives in RAID arrays. This is because if you have different sizes, such as 8Gb, 10Gb and 11Gb, you are restricted by the smalles disk to 24Gb (8Gb*3.) Also, a drive which is slower than the rest will act as a bottleneck, so performance will not improve as a result.
RAID Level 1 is also known as mirroring. An exact copy of what is written to one drive, is written to the other. This means, if one drive fails, there is still another drive to work from. You can work with this one drive until your failure gets replaced. The RAID 1 array then rebuilds itself, and you have a mirror drive again. This is quite simple. This is all done in real time. So as a file is written, it get's mirrored.
The bad things about RAID 1 are: It is expensive. You get half the space you pay for. If you want 40Gb of storage on RAID 1 you need 2 40Gb drives. The other bad thing is that there is no performance improvement.
The good thing with RAID 1 is that your computer will still work if one drive fails.
As with RAID 0, try to keep drives the same.
You can't just put a couple of drives into a computer and get them to RAID up. You need a RAID conroller. This is a piece of hardware which controls the operations required for RAID. A RAID controller calucates the file chunks to be sent to each drive in RAID 0, and sends the data to both drives for RAID 1. The fact that a conroller calcuates all this using hardware means your computer's processor isn't tied up working it all out.
Most RAID conrollers support RAID 1 and RAID 0. You can get more expensive ones which offer support for other RAID levels and contain large memory for write and read buffering.
Some new motherboards have RAID controller's build on board. You need no special connectors for your hard drives to use RAID. They still interface using IDE.
That is about as far as I can go without getting device specific or overly complicated. If you know very little about storage and computers in general this may wash over you. However, that means you probably aren't intending a RAID array yourself!
I've learned about RAID from loads of sources over the last three or so years. I still don't grasp it all, but I hope what I know is of use to other people.
A good article to read would Hardware Central
article. It explains in detail a lot about RAID, and at the same time reviews an inexpensive RAID contoller.
The above was taken from hdware-guide.com