ARGH! I hit <TAB> at the wrong time and submitted while I was still editing, making it look like my medication suddenly wore off. Please ignore above post.
RAID (Redundant Array of Inexpensive/Identical Disks) is relatively new to personal computers, though it’s proven its worth in servers and whatnot. I’ve just installed a RAID motherboard in my own machine; the Soyo Dragon+. The RAID capability was a major factor in choosing this particular board.
I’m not sure how much background I should give, but you described yourself as a “novice”, so here goes:
For microcomputers, hard drive interfaces have boiled down to 2 seperate yet proven approaches: SCSI and IDE. The latter is far more common, with motherboards of the last 10 years having the necessary support built right in. SCSI requires additional equipment, typically an extra card, unless you were willing to spend the big bucks and get a motherboard with an integrated SCSI chip.
An IDE “channel” can only have 2 devices hooked up to it, designated “Master” and “Slave”. These devices are almost always hard drives or CD/DVD drives. SCSI is more flexible. It can have (typically) up to 7 devices attached, covering a wider range of functions, including scanners, printers, etc.
When corporate computers act as file servers (i.e. the computers that stay in a locked air-conditioned room where only the nerds can get at it), hooking up several identical hard drives to a single SCSI channel proved to be highly desirable. There are several RAID configurations, the two most common of which are:
RAID 0: Two or more drives of the same size are “striped”, that is, a chunk of data is broken up so part gets written on sector X of drive 1, part gets written on sector X of drive 2, part gets written on sectorX* on drive n, etc. Ideally, all the drives would be read at the same time and the retreival of the data would be much faster than if all of it was stored on a single drive. Three 10-gig drives would be viewed as a single 30-gig drive. The main advantage was speed, often several times faster than normal performance. The main disadvantage is that if one drive in the array failed, you were screwed, because your data would have this big gap in it.
RAID 1: Two drives of the same size are “mirrored”, meaning the information that gets written to one also gets written to the other. Main advantage: security. If one drive failed, your data was still intact. On many servers, you could even remove the malfunctioning disk and replace it while the machine was still on (“hot-swapping”). Down time is minimal. The main disadvatange is cost, since two 10-gig drives are effectivley acting as only one 10-gig drive.
There are other RAID configurations that involve combinations of stirping and mirroring, and/or different numbers of drives, etc.
SCSI had a lock on RAID until some manufacturers started producing IDE RAID cards. You could pop one of these into your computer, hook up two identical (and relatively inexpensive) IDE hard drives to it, and set up a RAID 0 or 1. SCSI drives and cards were (and still are) expensive and since scanners and other equipment have shifted to USB, the technological advantage of SCSI over IDE has narrowed greatly. Lately, some motherboard manufacturers have started integrating these IDE RAID chips, so you need not use up a PCI slot, reducing the advantage even more. The March 2002 issue of Maximum PC contains an article comparing a top-of-the-line SCSI drive with its IDE counterpart and concludes that while SCSI still has its place in servers, they’re overkill and overpriced for desktops. IDE won the contest.
The Soyo Drgaon+ seems typical for an integrated motherboard. I have a total of four IDE channels: IDE1, IDE2, IDE3 and IDE4. Of these, IDE3 and IDE4 are controlled by an onboard Promise IDE RAID chip. IDE1 and IDE2 remain conventional interfaces. The arrangement I have chosen is as follows:
IDE1- Master:Quantum Fireball 20GIG Slave: none
IDE2- Master:Samsung CD-RW Slave: Phillips DVD
IDE3- Master:Western Digital 40GIG Slave:none
IDE4- Master:Western Digital 40GIG Slave:none
CD-drives apparantly can’t be put on the Promise RAID interface but I could have chosen to treat IDE3 and IDE4 like normal interfaces and had 8 independent IDE devices. Changing this is a matter of a jumper, a BIOS change and some software. Not a big deal.
When I bought this board, I also bought a 40GIG drive identical to the Western Digital already in my possession. You can set up RAIDs with non-identical drives, but somewhere along the way you’ll be wasting capacity. I plan to use this system for video editing, which requires heavy-duty drive use. The testing software that came with my video capture card reports the following speed tests:
C-Drive (my 20GIG non-RAID drive)
Writes 25.65 MB/sec with a low of 3,2 and a high of 60.8
Reads 20.08 with a low of 2.7 and a high of 32.3
V-Drive (my 80GIG striped drive, consisting of two 40GIGs mashed together)
Writes 61.41MB/sec, with a low of 10.4 and a high of 73.9
Reads 56.11, with a low of 6.4 and a high of 68.8
The performance of the RAID on average is more than twice that of the C-drive alone. There are a few minor factors to consider, such as the C-drive containing my WinXP files, etc, but those are fairly trivial.
If you plan to write a lot of CDs, though, you don’t really need RAID at all. Even a slowpoke hard drive will easily outpace a snappy CD-burner. The factors to look for on a burner are high write speed (obviously) and BURN (Buffer UnderRuN) Proofing. This feature allows CD-writers a bit of flexibility and make it far less likely that you will pile up a stack of useless “coasters” because somebody in your house flicked a light-switch and made the computer hiccup.