Fundamentals of RAID
A couple of the recent Technical Guidelines from http://www.geeks.com have created the reference to RAID, but the level of information needed in those tips did not shed much light on what RAID actually is. The number of e-mail reactions and feedback in the Visitors Digress area was effective enough that a release to basic principles of RAID would be an appropriate Technical Tip, so here it is.
The word RAID seems like it might explain something Marine corps perform in Fallujah or a can of what all pests worry, but it is actually an abbreviation that appears for Repetitive Range of Separate (or Inexpensive) Drives. Based upon on who you talk to, the correspondence “I” can stand for either independent or affordable, but in my viewpoint, independent is more appropriate, and far less very subjective.
RAID usually allows information to be published to several difficult drives so that a failure of any one generate in the range does not lead to the lack of information, as well as increasing the human body’s mistake patience. I say RAID usually does this, as there are several RAID options offering different techniques to redundancy, but some RAID options are not redundant at all. Fault patience represents a human body’s ability to carry on working when provided with a component (or software) failing, as should be experienced when a difficult drive generates is not able in one of the redundant options of RAID.
The primary components needed to run RAID has a set of printed difficult hard pushes and a RAID operator.
RAID can be run on any kind of difficult drive generate, such as SCSI, SATA, and ATA. The amount of difficult hard pushes needed depends on the particular RAID settings selected, as described later. I discuss the need for printed difficult hard pushes, and although this is not absolutely necessary, it is suggested. Most arrays will only be able to use the potential of the tiniest generate, so if a 250GB Hitachi generate is added to a RAID settings with an 80GB Hitachi generate, that additional 170GB would probably go to waste (the only time that this doesn’t apply is in a RAID settings called JBOD (Just a Collection Of Disks); which really “isn’t a RAID configuration” but just a practical factor that a RAID operator can do – see “Basic RAID Configurations” below for more information). In accessory for related capabilities, it might be of interest that pushes coordinate in terms of rate and exchange rate as the efficiency of the range would be limited by the most fragile generate used. One more area that should be considered while related is the kind of difficult drive generate. RAID remotes are usually for either SCSI, SATA, or ATA specifically, although some techniques allow RAID arrays to be managed across remotes of different types.
The RAID operator is where the information wires from the difficult hard pushes are linked, and performs all of the handlings of the information, like the common generate relationships available on a mother board. RAID remotes are available as add on bank cards, such as this Rubber Picture PCI ATA RAID operator, or incorporated into motherboards, such as the SATA RAID operator available on the Asus K8V SE Luxurious (http://www.geeks.com/details.asp?invtid=K8VSE-DELUXE). Motherboards that include RAID remotes can be managed without the use of RAID, but the incorporation is a nice feature to have if RAID is considered. Even for techniques without on board RAID, the relatively low cost of add on bank cards makes this part of the update relatively comfortable.
Another piece of components that are not needed, but may are great in a RAID range is a hot-swappable generate bay. It allows an unsuccessful difficult drive generate to be taken off a live program through opening the bay and moving the generate crate out of the situation. A new generate can then be slid in, closed into place, and the program won’t miss a defeat. This is commonly seen on SCSI RAID arrays, but some IDE RAIDS bank cards will also allow this (such as this device produced by Guarantee Technology: [http://www.promise.com/product/product_detail_eng.asp?productId=92&familyId=7]).
RAID can be run on any modern os provided that the appropriate motorists are available from the RAID controller’s producer. A pc with the os and all of the application already set up on one generate can be easily be duplicated to other people generate by using the application like Norton Phantom. But it is not as easy when going to RAID, as a customer who wants to have their current program with only one bootable difficult drive generate improved to RAID must start from the start. This means that the os and all application needs to be re-installed from the beginning, and all key information must be supported up to be renewed on the new RAID range.
If a RAID range is preferred in a process for use as storage, but not as the location for the os, things get much easier. The current difficult drive generate can remain unchanged, and the necessary settings can be generated to add the RAID range without starting from the beginning.
Basic RAID Configurations
There are about twelve different types of RAID that I know of, and I will explain five of the more common options, and usually provided on RAID operator bank cards.
RAID 0 is one of the options that does not offer redundancy, which makes it probably not a real RAID range. Using at least two disks, RAID 0 creates information to the two pushes in a changing style, known to as striping. If you had 8 sections of information, for example, amount 1, 3, 5, and 7 would be published to the first generate, and amount 2, 4, 6, and 8 would be published to the second generate, but all in successive order. This process of breaking the information across pushes allows for a theoretical efficiency increase of up to dual the rate of only one difficult drive generate, but real life outcomes will usually not be nearly so great. Since information is not published to each hard drive, the failure of any one generate in the range usually outcomes in a complete information reduction. RAID 0 is suitable for people who need to access large data files quickly, or just demand top rated across the board (i.e. game playing systems). The potential of a RAID 0 range is equivalent to the sum of the personal pushes. So, if two 160GB Seagate pushes were in a RAID 0 range, the complete potential would be 320GB.
RAID 1 is one of the simplest arrays that provide redundancy. Using at least two difficult hard pushes, information is published to both pushes in a method known to as reflecting. Each drive’s material is similar to each other, so if one generates is not able, the program could proceed to work on the staying excellent generate, which makes it an ideal choice for those who value their information. There is no efficiency increase as in RAID 0, and in fact, there may be a minor reduce compared to only one generate program as the information are prepared and published to both pushes. The potential of a RAID 1 range is equivalent to 50 percent the potential of the sum of person pushes. Using those same two 160GB Seagate pushes from above in RAID 1 would lead to a complete potential of 160GB.
RAID 0+1, as the name may indicate, is a mixture of RAID 0 and RAID 1. You have the best of both planets, the efficiency increase of RAID 0 and the redundancy of RAID 1. At the least four pushes is needed to apply RAID 0+1, where information is published in both a replicated and candy-striped style to the four pushes. Using the 8 sections of information from the example above, the written design would be something like this… Chunks 1, 3, 5, and 7 would be published to pushes one and three, and sections 2, 4, 6, and 8 would be published to pushes two and four, again in a successive manner. If one generates should don’t succeed, the program information is still unchanged. The potential of a RAID 0+1 range is equivalent to 50 percent the complete potential of the personal pushes. So, using four of the 160 GB Seagate pushes outcomes in a complete potential of 320GB when designed in RAID 0+1.
RAID 5 may be the most highly effective RAID settings for the common customer, with three (or five) disks needed. Details are candy-striped across all pushes in the range, and moreover, equality information is candy-striped as well. This equality information is actually a check on the information being published, so even though the information is not being published to all the pushes in the range, the equality information can be used to rebuild a lost generate in the situation of failing. Perhaps a bit difficult to explain, so let’s turn to return to the example of the 8 sections of information now being published to 3 pushes in a RAID 5 range. Chunks one and two would be published to generate one and two respectively, with a corresponding equality amount being published to generate three. Chunks three and four would then be published to pushes one and three respectively, with the corresponding equality amount being published to generate two. Chunks five and six would be published to pushes two and three, with the corresponding equality amount being published to generate one. Chunks seven and eight take us returning to the start with the information being published to pushes one and two, and the equality amount being published to generate three. It might not appear to be it, but due to the equality information being published to the generate not containing that specific pieces of information, there is full redundancy. The potential of a RAID 5 range is equivalent to the sum of the capabilities of all the pushes used, less one generate. So, using three of the 160GB Seagate pushes, the complete potential is 320GB when designed in RAID 5.
JBOD is another non-redundant setting, which does not really offer a real RAID range. JBOD appears for Just a Collection Of Drives (or Drives), and that is actually all that it is. RAID remotes that support JBOD allow customers to neglect the RAID features available and connect pushes as they would to an ordinary generate operator. No redundancy, no efficiency increase, just additional relationships for including more pushes to a process. A smart factor that JBOD does is that it can cure the odd scaled pushes as if they are only one quantity (thus a 10GB generate and a 30GB would be seen as only one 40GB drive), so it’s great to use if you have a whole lot of odd scaled pushes seated around – but otherwise it is better to go with a RAID 0, 1 or 0+1 settings to get the efficiency increase, redundancy or both.
Implementing RAID may seem overwhelming those new to the idea, but with some of the simpler options, it is not much more involved than establishing up a pc to use an ordinary generate operator. But, the benefits of RAID over only one generate program far over-shadow the other concern needed during set up. Losing information once due to difficult drive generate failing may be all that is needed to persuade anyone that RAID is right for them, but why to wait until that happens.