We recently produced a quick animated video showing the barebones basics of RAID. If you missed it, you might want to take a look now. This article aims to quickly recap that video and take you on a slightly deeper dive into which RAID levels might be best for your needs and how to implement them on your G-Technology storage enclosure. By the time we’re done, you’re going to know:
- How different RAID levels are suited to different workflows
- How switching between RAID levels is probably easier than you think
- How to change RAID levels on the fly with a G-Technology storage enclosure
The Web is full of geekspeak about RAID (including this comprehensive Wikipedia article), so we’re only going to skim over the basics as they relate to creative workflows.
RAID 0 (striping)
A hard drive or SSD is basically a parking lot for data. Cars (data blocks) come in, park for a while, and often go out again. With one drive in play, you have a single parking lot. And you know what happens when a lot of cars try to enter a parking lot at once: congestion. Everything bottlenecks at the entrance.
RAID 0 opens more parking lots to accept that traffic. If you have two drives, then you have two parking lots available, thereby cutting the congestion at each entrance nearly in half. (Technically, the congestion is caused more by cars pulling into their spots slowly, and thus blocking traffic, but we’re simplifying here.) The more drives you have in a RAID 0, the faster the total performance, although there are diminishing returns with each addition.
Warning: The car analogy falters a bit because you’re not parking whole files on a drive. With two drives, it’s more like you’re parking half a car in one lot and the other half in the other lot, and they get rejoined when they pull out onto the road. (This process is called striping.) Now, if one drive fails — say, an asteroid falls on one parking lot and obliterates its contents — you’re never going to have a whole car again, right? Thus, RAID 0 is fast, but it’s risky. If you lose even one drive from a RAID 0, all data is permanently destroyed.
We recommend RAID 0 in situations where the fastest possible throughput can assist with large data transfer completions and real-time editing of very high-resolution content, but only if you already have that data safely backed up…preferably twice.
RAID 1 (mirroring)
Drives fail for all kinds of reasons, including cosmic particles, extreme wear, and cats. Obviously, this is why we back up data. You want that protection process to be as fast and brainless as possible. The more you have to interact with the backup operation, the higher the chance of mishap and human error.
RAID 1 simply says, “Oh, you’re writing something to that drive there? I’ll just make an instant copy over here.” This way, if one drive fails, you have the other functionally identical drive(s) up, running, and ready to step into service with zero downtime.
Naturally, if one drive fails in a two-drive RAID 1, you’re down to having one drive, and if that one fails, your data is toast. Hence, you want to replace failed drives ASAP. Also understand that the “mirrored” drive(s) will be invisible to you, so if you put two 8TB drives in a RAID 1, you have 16TB of raw capacity but only 8TB of usable capacity.
What happens if you have a two-drive RAID 0 (striped), and then you mirror (with RAID 1) that array onto another pair of drives? You’d have a mirrored RAID 0, better known as a RAID 0+1, also called RAID 01. You can also have a striping (RAID 0) of mirrored (RAID 1) drives, dubbed a RAID 1+0, or RAID 10. These approaches minimize the risk of RAID 0, but they still are costly in terms of lost capacity. The following options tend to be more common and preferable.
The parity computation process that underlies RAID 5 is, for the average layperson, complex and dull. So forget all that. Think of it this way:
You’re one of three Neanderthals around a campfire. The three of you have been entrusted with the tribe’s Best-Ever Hunting Story. Unfortunately, none of you has enough brain power to remember the whole tale, so you each retain pieces of it. In addition, you each have distributed parity. This means that you have enough bits of information in your head about the others’ parts of the story to enable you to reconstruct their part of the tale. Thus, even if one of you is smooshed by a woolly mammoth or eaten by a scimitar cat, the Best-Ever Hunting Story will live on, and you can then “reconstruct” the missing part of the story and teach it to the new third Neanderthal when he or she joins the group.
RAID 5 requires at least three drives. It stripes data, including parity data, across every drive in the array, but this speed benefit is somewhat offset by the extra work required for parity handling. It is an “N-1” arrangement, which means that you get the raw capacity of however many drives are in the array (N) minus one drive. RAID 5 tends to be the default mode in storage enclosures, as it offers a solid compromise between speed, protection, and capacity.
Like the previous “nested” levels, RAID 50 is a RAID of RAIDs. In this case, it involves striping (RAID 0) across at least two RAID 5s. As such, this level requires at least six drives. In practice, RAID 6 (below) is generally used instead of RAID 50.
RAID 6 is essentially the same thing as RAID 5, only it’s N-2 and requires at least four drives. This way, if two drives in the array fail, which can and does happen, the array’s data remains intact. We recommend RAID 6 for situations in which data protection takes a somewhat higher priority, such as when high-intensity editing must be done in the field, but you haven’t had the time or resources to make adequate secondary backups. In conjunction with this, be aware that when a drive fails in a RAID 5, the remaining drives operate under a higher strain level, which can lead to premature failures if left unresolved. Hence, RAID 6 becomes even more attractive if spare drives are going to be in sparse supply, such as when shooting in remote locations. Because RAID 6 is N-2, a collection of six 10TB hard drives would yield 40TB of usable capacity.
A Few Clicks to Optimized Storage
In short, if you only have two drives, as in a G-RAID with Thunderbolt 3 or a G-SPEED Shuttle with ev Series Bay Adapters, then RAID 0 makes sense if you need top performance for editing. RAID 1 will likely prove smarter if you need more protection and can give up some capacity. Once you step it up to three or four drives (examples: the four-bay G-SPEED Shuttle or the eight-bay G-SPEED Shuttle XL, both with Thunderbolt 3), RAID 0 and 1 remain viable options, but most users will opt for RAID 5 or, if the capacity of two drives can be spared, RAID 6.
You understand that different RAID levels are appropriate to different drive counts and storage priorities. Similarly, you may find it advantageous to change RAID levels in the same enclosure as your needs for that enclosure change. For example, say you have an eight-bay G-SPEED Shuttle SSD purchased for editing on a multi-camera 8K project. You’re generating over a terabyte of footage per day on a tight deadline, and editing speed is paramount. RAID 0 might be appropriate if you have a couple of moderate-capacity Thunderbolt 3 enclosures in play that won’t slow down the workflow. If you need to balance protection and speed, then the default RAID 5 is likely a good choice. If your career hangs on the content of this box and no other backups can be brought online, then definitely go for RAID 6.
Here’s the thing that many people overlook: Jobs finish. Priorities change. You might want to change that RAID 5 default to RAID 0 for a week, and then bump it into RAID 6 when taking that enclosure overseas. This is easy to do! Let’s follow the few steps.
Bring up the G-Technology management application for the storage enclosure. In this case, we’re using the app for the G-SPEED Shuttle SSD. Along the top, click the icon for the Logical Drive tab.
Our model comes configured with eight 2TB SSDs. This gives a raw capacity of 16TB, but because the default level of RAID 5 is N-1, we’re left with a usable capacity of 14TB. To change RAID levels, we must first delete the current logical drive (which shows up in your drive listing as a single-letter volume), and then create a new logical drive in its place. In most cases, THIS STEP WILL DESTROY ALL DATA ON THE DRIVE, so be sure to back up your data onto separate storage. When you’re ready to proceed, select the array, and then click the Delete button.
To make sure you don’t accidentally delete your data, the storage software will prompt you to type the word CONFIRM, and then click the Confirm button.
Boom! No more drives. In the top-right corner, click the Create Logical Drive button.
The software finds that you have 16TB of free drive capacity available. Click the Next button.
In the RAID Level field, select the level you wish. In the example here, we’ve selected RAID 6. Your logical drive needs an alias, so after much thought, we arrived at the highly creative name of “RAID 6.” For the sake of simplicity in this example, we left all other options unchanged. Hit the Add button and you’ll see a summary of your new volume appear under New Logical Drives on the right.
And that’s it! You can get fancier with options and creating multiple RAIDs within a single enclosure, but that’s another tutorial for another day. For now, just know that you can optimize your storage workflow by changing your enclosure’s RAID settings, and changing those settings could hardly be easier. It can literally be done in one minute. Why not take advantage of this powerful, flexible capability?