A breakdown on common RAID configurations

What’s a RAID?

A Redundant Array of Independent Disks is a way to make a computer treat a number of physically distinct drives as one big drive. There are several different RAID configurations, each with advantages and disadvantages. The most common types of RAID are 0, 1, and 5. Here are the configurations mentioned in this review.

In this GCN Lab comparison report:

NAS appliances cover the middle ground of extra storage
What Is A RAID?
Buffalo TeraStation III
LaCie 5big Network
Sans Digital EliteNAS
Seagate BlackArmor NAS 440
WD ShareSpace
Gaining Virtual V-locity

JBOD span: Not usually considered a RAID configuration, JBOD stands for “just a bunch of disks” and is exactly that. In this mode, disks of equal or varying sizes are linked and treated as one big, single drive. This configuration has the advantage of being able to use drives of different sizes and utilizing nearly all their combined capacity. However, a JBOD span has no fault tolerance — if one drive fails the whole thing fails — which is why it isn’t used often.

RAID 0: This configuration uses striping, which is one of two basic RAID techniques. Each disk's space is split into stripes of a certain length, which can vary based on different RAID management programs but is always the same within a single RAID. The software then writes to each stripe before moving onto a stripe in the same position on the next drive. So the first stripe on disk one would be followed by the first stripe on disk two, and so on. This configuration allows you to use all the drives' capacity. But with RAID 0, if a single drive fails, the entire RAID fails, and you will likely lose all your data. It does offer better disk access times than JBOD.

RAID 1: This RAID uses the other basic technique, called mirroring. The disks are paired, and the second disk in each pair is an exact copy of the first. Each pair is treated as a separate drive. This has maximum redundancy, as both drives in a pair would need to fail for that logical drive to fail. Also, RAID 1 tends to have faster access times than other configurations. However, the capacity is half that of RAID 0.

RAID 1+0 (or 10): This approach combines the techniques of RAIDs 0 and 1. Each pair of drives is mirrored, and each mirrored pair is striped so that the RAID is treated as one drive. This is just as secure as RAID 1 but has half the capacity. The difference is that each pair is not treated as a separate drive but one contiguous one.

RAID 5: This more secure RAID uses striping but does something different with the stripes. For each set of stripes — such as the first stripe of every disk in the RAID — one of the disk's stripes contains parity data corresponding to the data on the other disks. The disk that has the parity information alternates with each set of stripes. The advantage is that if one disk fails, the RAID continues to function, and the failed drive can be rebuilt onto a replacement from the information on the other disks. Still, RAID 5 isn't 100 percent bulletproof in protecting data, only the mirroring of RAID 1 can offer that. But RAID 5 is highly reliable. The advantage it has over RAID 1 is increased capacity. With only one disk in the array being used for storing parity, the capacity of a four-disk array would be the total size of the remaining three disks.

RAID 6: More secure than RAID 5 at the cost of some drive space, RAID 6 uses two drives for the parity of each stripe. The two drives used for parity alternate with each stripe. For instance, if drives 3 and 4 in a four-disk array have the parity for the first stripe, then drives 2 and 3 might have it for the next stripe, and so on. This has the increased advantage that data would remain intact even if two drives fail. Of course, this means that there is one less drive’s worth of storage capacity, as the equivalent of two of the drives are used for parity. It’s often only available on high-end RAID configurations.

About the Author

Greg Crowe is a former GCN staff writer who covered mobile technology.

inside gcn

  • Students in a Virginia Tech laboratory test Popcorn Linux -- an operating system that can compile different programming languages into a single format. Credit: Dr. Binoy Ravindran

    Popcorn Linux OS gives processors a common language

Reader Comments

Thu, Dec 3, 2009

I believe that the comment on Raid 10 having half the capacity of Raid 1 is inaccurate. The "kick" against Raid10 is that it requires 4 disks which puts it in competition with Raid 5 and/or 6 rather than Raid 1. It has the same capacity as Raid 1. If you have 4 1TB disks you can create a 2 TB Raid 1 drive with the mirror using the other two 2 TB drives. In Raid 10 you end up with the same 2 TB drive but the data is striped over the two drives making the data. EG. Drives 1 and 2 are mirrored and drives 3 and 4 are mirrored. Data is striped across drive 1 and 3 and due to the mirroring is also striped across 2 and 4. Raid 10 is faster than Raid 1 and slower than Raid 0 but gives you data reliability. With it you use 4 TB of space to get 2 TB of storage. With Raid 5 you would get 3 TB of storage using the same 4 TB of drive space. I am not sure which would have the faster access though (I suspect Raid 5). Raid 6 with four drives gives the same capacity as Raid 10 but may have a performance edge. Where it has the best advantage is that any two drives can fail and you still have your data. With Raid 10 your data may be safe with two drives failing but you do run the chance of losing your data if the wrong two drives fail. If both drives in a mirror fail you lose everything. If 1 and 2 or 3 and 4 fail you lose all data. If 1 and 4, 2 and 3, 1 and 3 or 2 and 4 fail you still have your data. With Raid 6 you have 100% chance of data security with a two drive failure while Raid 10 gives only 67%. None of this covers the possibility of failure to rebuild due to an unrecoverable read error in the data being used to recreate the disk. All hard drives have a specification of how many reads before the unit has an unrecoverable read error (usually at the bit level but still one bad bit makes one byte bad which can be a critical byte). This wasn't a problem when we had disks that were hundreds of GB in size but now that disks are into the TB range the odds of an error being found during the recreation of the drive are getting better which means that the recovery will fail.

Please post your comments here. Comments are moderated, so they may not appear immediately after submitting. We will not post comments that we consider abusive or off-topic.

Please type the letters/numbers you see above

More from 1105 Public Sector Media Group