We can combine multiple drives together to optimize data throughput and provide redundancy. In this video, you’ll learn about the redundant array of independent disks (RAID), and an overview of RAID 0, RAID 1, RAID 5, RAID 6, and RAID 1+0.
We rely on our hard drives, SSDs, and other devices to store large amounts of information, and often, this information is very important. We don’t want to lose the data that we’re storing on these drives.
But of course, these hard drives are physical devices that are constantly moving. The platters are spinning, the actuator arms are moving, and if any one of those components is to fail, everything on that drive will be inaccessible.
Fortunately, there are ways that you could combine multiple drives together to create redundancy, so if you do lose a drive, you can be assured that all of your data will still remain available.
And as we go through this video where we talk about drive redundancy, please keep in mind that this RAID configuration we’re going to speak of is not a method of backup. If you do have an array that is using RAID to be able to provide this redundancy, please keep in mind that you also need to maintain a completely separate backup process.
RAID is an acronym that stands for Redundant Array of Independent Disks. You might also see this referred to as a Redundant Array of Inexpensive Disks.
There are many different ways to implement RAID, and we’ll look at some of the more popular ones in this video. Some of these methods provide a way to keep your data redundant even if you lose a drive. But there are some RAID methods where redundancy is not available, so we do need to know the difference between one and the other.
In this video, we’ll step through RAID 0– you might also hear this referred to as striping; RAID 1, which is mirroring; RAID 5, which is striping with a single-parity drive; we’ll also look at RAID 6, which is striping with two parity drives; and we’ll also look at nested RAID, which are two RAIDs combined together, specifically RAID 1 and RAID 0. Sometimes you’ll see this written as RAID 10, and it refers to a stripe of mirrors.
Let’s start with RAID 0, which we refer to as striping. RAID 0 has at least two physical drives, and we’ll take the information we’re saving to these drives and split it across both of these storage devices. For example, if you have a single file, you could break that file up into eight different parts, and you would store a part of that file onto each of these drives.
For example, you might see one drive with Block 1A, another drive with Block 2A, back to the first drive with Block 3A, and so on. RAID 0 is well known for its speed. Because you’re writing a little bit of information to multiple drives, you’re effectively able to do this faster than writing all of that information to a single drive.
The problem, of course, is that if you do lose one of these physical drives, you now have lost access to your data because half of what was there normally is no longer available. For that reason, we often think of RAID 0 as having zero redundancy.
RAID 1 is mirroring, and as the name implies, mirroring means that we’re going to have copies of information across multiple drives. For RAID 1, we need at least two drives– we have those two here on the screen, and you can see that anything that we have on Disk 0 also has an exact duplicate of that information on Disk 1.
This means that we need effectively twice as much storage space to be able to store exactly the same information. Having that mirror of duplicate information means that we are going to use a lot of storage space to be able to maintain that redundancy.
But one good thing about mirroring is that if you do lose a physical drive, the other physical drive containing an exact duplicate of all of that data continues to be available. Obviously, we will want to replace that drive as soon as possible and recreate the mirror, but while we’re in that process, all of our data continues to be available and we can continue to work normally.
In our video on memory, we talked about how parity can be used to identify errors or to correct errors. We take that same idea and move it into RAID to be able to provide redundancy if we lose a physical drive. One of the ways we can do this with parity is using RAID 5, or striping with parity.
The striping that we’re doing in RAID 5 is identical to the striping that we were doing with RAID 0. Take a file, cut it into small pieces, and distribute those pieces across multiple physical drives. However, the last drive, we’re not going to take a piece of the actual file, we’re going to take the parity of the information that we just stored.
This means in a RAID 5 array of four physical drives, three of those drives would have data, and the fourth drive would contain parity data.
You’ll also notice that parity is distributed across physical drives to make the recovery process more efficient. You also get efficiency in how much data you’re storing and how you’re storing it. Since you’re just storing parity data on a separate drive, you’re not having to duplicate all of the data on the system. That means that you’ll have more drive space to be able to store your data instead of having to make copies of your data.
If you do lose a physical drive with RAID 5, you can take the data that still exists, combine that with parity, and you’re able to recreate the data that’s no longer there. You’re able to effectively, in real time, still maintain all of the data as if it was still physically located on that drive.
This parity calculation does require a bit of CPU overhead, so there may be times when using RAID 5 in a recovery mode where there is a bit of a performance hit.
RAID 6 is very similar to RAID 5, except we’re adding another storage drive and adding an additional parity block. This means if we lose a single drive, we’re effectively running the same way as RAID 5. If we lose two drives, then we can recreate the lost data using our existing parity data.
This means with RAID 6, we could lose a total of two physical drives and still remain up and running, although in a degraded state. This means that we could lose two physical drives in a RAID 6 array and still have access to all of our data.
Of course, this does mean that we need a separate physical drive to store that additional parity data. And because we’re adding additional parity data, we, unfortunately, are not adding additional capacity when we add that extra drive and create a RAID 6 array.
RAID 10, or what we call RAID 1+0, is combining RAID 0 with RAID 1. RAID 0, obviously is striping, just like we can see here. We have three physical drives, we’ve taken a single file, we’ve split that file up into 12 different blocks, and we’ve put one block on each of these drives.
Because this is RAID 0, we have zero redundancy. So if we were to lose any of these drives, all of our data would no longer be accessible. With RAID 1+0, we need to add that RAID 1 functionality, or mirroring. So we’ll continue to keep our stripe, but we will mirror each striped set of drives. This means that we have separate individual drives that have a copy of the striped information.
This requires at least four drives, but it also means that we could lose multiple drives and still be up and running. For example, in this scenario, we could lose one of the drives from the first mirror, one of the drives from the second mirror, and one of the drives from the third mirror, and we would still be up and running because we would still contain three separate drives from each individual stripe.