There are many ways to combine our storage devices to increase performance or provide redundancy. In this video, you’ll learn about the most popular RAID types used in our storage systems.
<< Previous: An Overview of Storage DevicesNext: An Overview of CPU Socket Types >>
RAID is an abbreviation that stands for redundant array of independent disks. We used to actually call this a redundant array of inexpensive disks. These days it makes more sense to call these independent disks. There are many different RAID levels. We’re going to look at a few of those RAID levels in this video. Some of these, even though the name says redundant as part of RAID, some of these RAID levels are redundant and some are not. So you have to be very aware when you’re putting together a RAID type whether it’s one that provides redundancy or does not provide redundancy.
The RAID formats that we’ll look at in this video are some of the most popular. RAID 0 is striping. We will look at RAID 1, which is also called mirroring. We’ll look at RAID 5, which is a striping with parity. And finally we’ll look at a nested RAID, which is a RAID 10 or RAID 1+0. Which is effectively a stripe of mirrored drives.
If you’re configuring a RAID array inside of your computer you’re going to be using either a software based RAID or a hardware based RAID. A software based rate is one that is built into your operating system. It all runs on software. You don’t need any special hard drive controllers or any special hardware whatsoever. You only need to have a number of drives that the operating system can see. Because this is running in software, the performance of your RAID array can vary based on how busy the operating system is with other tasks and how much information is being written to or read from the RAID array.
With hardware based RAID we’ve taken the operating system completely out of the equation. In this particular case we’re adding a piece of hardware to our computer. This might be built on to the motherboard or it might be a separate adapter card like the one you see here. And then we connect our drives directly to this hardware. This hardware based RAID handles all of the RAID performance and all of the RAID management. The operating system doesn’t see anything relating to these RAID pieces, it’s simply talking to this one piece of hardware. Obviously, the performance with a hardware based RAID tends to be a little faster than what you might find in a software based RAID solution.
Now that you’ve decided to run your RAID array in software or in hardware we now get to decide what type of array we’d like to install. The first one you could consider is a RAID 0. We also call this striping. That’s because we take our information, we split it up into different pieces, and we layer it across multiple physical drives. In the case of RAID 0 we need at least two physical drives to be able to do this. RAID 0 tends to have very good performance because you’re splitting the load across these multiple physical disks. But of course, because we’re splitting the data, if we lose any one of these physical drives we’ve lost half of our information and there’s no redundancy available. That’s why when you think of RAID 0 you should also think of zero redundancy.
Another RAID type you might want to consider is RAID 1. This is also called mirroring. And we call it mirroring because we’re creating an exact duplicate of your data. So we’re going to need at least two physical drives to do this. Instead of splitting your information between the different drives we’re putting an exact duplicate across these drives. As you can imagine this means that you are using a lot of drive space. You’re using twice as much drive space than you normally would because you’re storing the information twice across these separate physical drives.
Every single file and every bit of information is duplicated across both of those physical disks. And obviously this means you’re absolutely redundant between these two. If we were to lose any one of these drives the other drive has an exact duplicate of the information. So you can always be assured to be up and running with the data that you’re storing on this array. This also means that it’s easy to recover when a drive goes bad because you can simply remove the bad drive, replace it with a brand new drive, and it’s going to recopy everything over and create a new mirror drive array.
RAID 5 is a RAID format that takes the striping that we did with RAID 0 but adds a new piece of information called a parity bit to our data. We’re striping this information across multiple disks and we need at least three physical disks to be able to use RAID 5. But on the last disk we add a little bit of parity information. This means that we’re able to use the disk space efficiently. We’re not creating an exact duplicate of your data. We’re simply adding a little bit of parity information along to your data.
The advantage here is that if you do lose any one of these physical disks, we’re able to recreate the data because we have the parity information. This means that we have to perform calculations and figure out what the missing data is while these drives are in this limited state. But once you replace the bad drive it will rebuild that drive and everything will go back to the normal RAID 5 configuration.
The advantage of RAID 0 was very fast performance but we had the limitation of no redundancy. Well to get around that we combine RAID 0 with RAID 1 to come up with RAID 1+0. Which is effectively still using striping but we’re striping to a set of mirrored drives. To be able to do this, we need at least four drives. So we’re combining a lot of different capabilities here.
And you can see here I’ve got three sets of mirror drives here at the bottom. And I’m going to stripe between those three sets. If I lose any physical drive I’m still up and running because I have redundancy with the stripe that I’ve created. I’ve created all of these RAID 1 mirrors so that I’m able to do RAID 0 across all of them with a RAID 1+0.