It seems that every-one loves posting polls... so how about another discussion instead. In discussions on things like religion, everyone has an opinion because everyone is their own expert and it is based on a lot of opinion.
How about something that everyone isn't an expert in. Something that relates to computers that may require a little research, that will help us to learn more...
This discussion is on RAID devices. Do your homework.
What is RAID and what is the best type of RAID for an application that you would most likely use and why?
We use RAID 5 for a 500 Gig virtual disk that provides video recording and assesment storage. With this equipment, we needed the ability to replace a bad drive and have it rewritten to retain our video recording data.
RAID 5 utilizes parity and striping over multiple drives and has excellent read performance. For us, it was the ideal config to suit our needs.
Well, I thought I knew what RAID was, but maybe I could be enlightened.... I was under the assumption that RAID was when drives were hooked togethr to make one uber-drive, correct? So, what's RAID 5 mean?
Govt: RAID levels
If you want all out speed, go RAID 0 with a pair of ATA133 hard drives. Oh yeah, pray to god that one doesn't fail, cause if it does, what ever is on those two drives is gone...
Half you data is stored on each == 2x (~) data access speed. I beleive you can stripe across any number of discs (>1)
Does RAID still stand for Redundant Array of Inexpensive Discs?
I think the Inexpensive part is a bit of a misnomer.
I've always liked RAID 5 myself, but the powers that be where I am currently working have something of a problem with accepting it. They are using a hardware striping controller, but volume shadowing is done on the host.
The acronym RAID is most often decoded as "Redundent Array of Inexpensive Disks", (most often - the "I" is contentious). It is a method of controlling disks. It comes in several varieties, 0,1 and 5 being the commonest I guess. (3 to me resembles a poor mans 5).
0 is striping, that means spreading data from individual files accross several disks, it speeds things up, i.e. several heads are reading/writing at the same time, (not unlike interleaving memory).
1 is mirroring, writing the same thing on more than one disk. Good for reliability.
3 is like 0, but reserves a disk for hamming. It is a performance boost and you do get some reliability increase.
5 is super 0. Here the data is striped right down to the byte level, and in the event of a disk failure, you simply replace that drive and the RAID set will rebuild it, and all the time just keep running and running.
Whoops - a dose of concurrency methinks!!!
RAID now stands for Redundant Array of Independant Disks. (In 1987-88 it stood for Redundant Array of Inexpensive Disks... but was later changed).
It is an array of disks that makes a larger virtual disk. There are many different types of RAID...
Some of the more common...
RAID 0 (not a true RAID... Why?) = striping only
RAID 1 = mirroring
RAID 3 = data striping w/ a parity drive
RAID 5 = data and parity striping
RAID 0+1 = mirroring and striping
I can pull a drive out of my RAID 5 config and it still runs as if the drive was there (minus some performance and the fact that the drive isn't there so actual storage is lower), and when I put a new drive in... it is completely rebuilt as an exact of the drive that was removed.
How do you think that parity does this...?
Thanks for the info an link, guys.
Couple questions - what in the heck is hamming?
In RAID 5, when the disk fails, you say you just replace the new one and the RAID keeps running and rebuilds it. How does the RAID know what data was on the lost disk?
> RAID 0 (not a true RAID... Why?)
According to the link drewpee gave, because the Berkeley engineers that created RAID didn't make a RAID 0.
> I can pull a drive out of my RAID 5 config and it still runs as if the drive was there (minus some performance and the fact that the drive isn't there so actual storage is lower), and when I put a new drive in... it is completely rebuilt as an exact of the drive that was removed.
Whoa, how does this happen?
> more concurency methinks
On an extreme scale...
>>> what in the heck is hamming?
Sorry cheez. The extra disk is used for storing error correction information, whether it be parity or whatever. I always tend to think of hamming codes, (search google), when I think of error detection and correction because about 15 years ago I had to design a suite of that kind of stuff! If you're interested, I recommend you have a look into it, it will explain a lot of miraculous error corrections!
>>> In 1987-88 it stood for
Oops, showing a bit of age there aren't I. I didn't know it had been officially changed. I have recently seen "Interchangeable" for the "I".
I will step out for a few Govt. Someone else can answer your question, or you can find out why yourself if you like. If when I get back from my break, nobody answered... I will type something up that describes RAID 5.
The bold parity bits are striped across the drives. You will notice that the parity is an AND function of the other three drives. Utilizing a configuration like this, you can remove a drive and still tell what information was on the drive
By reading across A, B, and C... you can formulate what was on drive D... so drive D doesn't need to be there to keep running, and if you put a new drive in for D, it can be rebuilt to look like the old drive.
> Make sense?
Except - if you pulled out more than 1 drive at a time, you'd have a 50-50 chance of getting the wrong data back, right? (assuming you can even do this)