Raid 0

I’m about to set up a RAID 0 between 2 seagate 120g SATA drives on an Asus a7v8x deluxe. I can’t find any documentation on whether or not this is practical or if it even works. Does it matter if I stripe on the same SATA controller? What is the best way to do this with the hardware mentioned above.

The board supportd RAID 0/1 but between which controllers? Will any combination work? I wouldn’t want to stripe between SATA and ATA 133.

I’m a beginner so any info you could provide would be much appreciated.

Jongri

argle gargle to you too jongri

Well, this page seems to imply that the the RAID controller is only for the ATA133 devices, not the SATA connectors.

Go here and get yourself a manual.

You need a RAID controller that supports your mode.

After that said, using RAID 0 is just dumb.

No, it isn’t. It may not be necessary, but it isn’t dumb. If you plan to use your computer for heavy-duty video editing (as I do) or as a file server for noncritical information, than having a RAID 0 is a great way to get a speed boost with minimal expense.

If you want speed and security, I’d suggest setting up a RAID 10 with four drives. So far, that’s more that I need and just doing it for the hell of it would be, I admit, pretty dumb.

Could someone explain what the hell Raid 0 is for a non-software literate person? I’m not trying to be derisive, I’m just curious.

Okay, RAID stands for Redundant Array of Inexpensive Disks (or something similar). It’s not a software thing at all; it’s a means of arranging hard drives in such a way that:
[ul][li]The information is broken up or “striped” across several hard drives. Each “read” operation takes time, so reading 2 units of data will take 2 “read cycles” (speaking simply). If you break up your data so that each of two hard drives has one of the data units, you can read them simultaneously, taking only one “read cycle”. Your two hard drives are treated as one big hard drive, which can be read from (and written to) at twice normal speed. This is known as a RAID 0, which is actually a misnomer, since none of the information is really “redundant” (see next item). RAID 0 arrangements can be set up with three or even four drives, with corresponding (but gradually diminishing) improvements in speed.[/li][li]The information is copied (“mirrored”) across several drives, making redundant copies. If you absolutely need your information to be kept safe, it might be worthwhile to install a second hard drive and write the same information to both at the same time. If one drive fails, your machine automatically jumps to the other, and you can replace the defective drive, copy the contents from the working drive, and carry on with no major interruption in service. This is known as a RAID 1.[/li][li]Some combination of the two. RAID 10 (one-zero, not “ten”) uses four hard drives, essentially two identical striped pairs. There are other arrangements like RAID 3 and RAID 5 and whatnot that use various combinations of drives in an attempt to balance speed, safety and expense.[/li][/ul]

Until a few years ago, RAID configurations required SCSI drives, which are fast but expensive. Lately, though, setting up RAIDs with the more common and less expensive IDE hard drives has become possible. If you’re buying a generic (i.e. IDE) hard drive to upgrade your computer, you might consider getting two identical hard drives and a RAID card. Once installed, a RAID 0 stripe gives you high speed, but unless you’re using it for drive-heavy operations like video editing or file serving, it may be more expense, complexity and risk than you actually need. I use a single hard drive to store my Windows junk and main programs, and a pair of drives in a “stripe” (RAID 0) as the “scratch” disk for my video projects. I can capture several hours of video without losing a single frame because my RAID is fast enough to keep up, even at 30 frames/second at high quality. I couldn’t do that on my single “C:” drive. By the same token, I don’t bother installing programs like Microsoft Office on my special “G:” drive (as I said, the drives in the RAID are viewed as a single unit, to which I can assign any letter I want) because that would be a waste.

A RAID0 is risky because if either drive fails, I’m screwed, so I only keep temporary or less-important files on the G: drive. When I finish a video project, I burn it on CD-ROM as soon as possible.

Tom’s Hardware Guide has an excellent introduction to IDE, with Part 2 and Part 3 getting into the nitty-gritty of RAID configuration.

RAID is an acronym, orginally meaning, if I remember correctly, “Redundant Array of Inexpensive Disks”. The idea of RAID is, rather than a single physical hard drive being the data storage device, a number of drives can be used together to provide performance or fault tolerance, although they only appear as one drive to the operating system. The various RAID levels are given numbers and the ones that you most commonly see are Level 0, 1, and 5. They work as follows:

RAID 0: Disk Striping. The factor limiting hard drive speed is the physical workings of the disks, not the electronic movement of data. By having multiple disks set up so that the computer can write a piece to disk one, then disk 2, etc, it can give you a significant increase in performance, although there is no fault tolerence (actually, since failure in either disk can make you lose your data, it is somewhat more risky than a single disk).

RAID 1: Disk Mirroring. This is simple. You have two hard drives. Any time you write data on one drive, it writes on the other one as well. If a drive fails, no problem - you still have the other drive with all the data on it.

RAID 5: Disk Striping with Parity: This is a bit more complex, but is used quite often for business servers. You need a minimum of 3 disks for this. Without making it too technical, when data is written, a little extra is written on all of the disks, so that if any one disk fails, it can be replaced and the data is then reconstructed from the “parity” data on the disks.

Hope this answers your questions.

It is dumb because you are not going to get much of a speed boost. Unless you purposely buy slower HDDs (e.g. DMA100, SCSI 3), there’s almost no benefit in running a RAID 0.

Suppose you hook two DMA133 HDD’s together, unless the controller is significantly faster, somewhere at around 180MB/s transfer rate, there’s almost no boost in speed after you have taken CPU overhead into account.

SCSI is better because it has almost no CPU overhead, but then, unless your mobo has a SCSI controller onboard, the PCI bus becomes bottleneck.

Unless you are using the newish 64-bit PCI bus and a card like the Adaptec 29160. Myself, I just use the IDE RAID controller on my MB for JBOD (no RAID at all)…

On a somewhat related tangent, here is a great article on using IDE RAID on a large scale (like 70 TB).

http://www.tomshardware.com/newsletter/vol3/16/70tb.html

RAID 0 also provides a very tangible benefit by simulating a much larger single drive with performance characteristics that are equal or superior to the component parts. The lack of redundancy actually makes the storage 100% efficient if identical drives are used.

The downside of course is that if one of those drives fails (and many of them are only coming with a one year warranty these days) the whole array is kiboshed.

Drive size can be important if, say, someone is amassing an enormous archive of pornographic video files on a single drive, something I happen to know jongri is interested in. Aren’t you, porno boy?

(And let me take this opportunity to introduce jongri to everyone here at the SDMB. He’s a nice enough sort, even if he is being slowly driven insane by living with me in the Playground of the Damned.)

I beg to differ. I have a 4-drive RAID 0 setup on my PC consisting of ATA133 7200 RPM drives, plus one more of teh same drives as a boot drive. On this system it takes an average of 10 seconds to write a 100MB file onto the stand-alone drive and 2.8 seconds to the RAID array. Read speed is about 3 sec vs. 1.5 sec. I call that a significant difference. My benchmark program writes a dozen of those files in succession so I don’t think it’s affected by cache and buffer. It’s a Linux software RAID, by the way, and only used to store downloaded data files.

Hmm, “downloaded data” might give you the wrong impression. The Linux box in question is my primary workstation at work, used to analyze satellite data which are available by ftp.

I have 2 DMA133 drives in a RAID 0 place and the speed benefit is apparant.

Next you’ll be saying “flying in airplanes is dumb! It can’t possibly be faster than walking.”