RAID Server Data Recovery
Pull the bays out of your server and you’ll see a lot of hard drives—anywhere from three or four (if you’ve got a small business) to one or more dozen drives. These hard drives are linked in a RAID array. RAID arrays are fault-tolerant, but not fault-immune. Sooner or later, your server will crash. If your server has crashed, our RAID server data recovery technicians can help you.
RAID Server Overview
Combining hard drives in a RAID array offers massive benefits to storage capacity, fault tolerance, and performance speeds. This makes them ideal for servers, which often have to provide a lot of services very quickly. There are quite a few different levels of RAID array, but only some see play in a typical server setup. RAID-0 simply stripes multiple hard drives together for massive improvements in read/write speeds and capacity. But it has no fault tolerance whatsoever. And it is more than twice as likely to fail as a single hard drive on its own. RAID 0 is a huge liability for a server environment, and the chances of anyone using it in a server are slim to none.
RAID-1 simply takes two hard drives and mirrors them, making them exact copies of each other. While this helps with redundancy and improves read speeds, it makes hard drive write speeds much slower. It also provides no storage capacity gains. For this reason, RAID 1 is rarely used in servers. It isn’t much better than having a single drive.
RAID 2, RAID 3, and RAID 4 exist, but are obsolete. The majority of modern RAID controller cards no longer even support these configurations. For the most part, RAID 3 and RAID 4 were just worse versions of RAID 5, and as such were eventually completely superseded by higher RAID levels.
You can read about our RAID data recovery methods in great detail here. Below are some of the RAID levels our data recovery experts tend to see when a client comes to us:
RAID 5 Server Recovery
The vast majority of servers our RAID server data recovery experts see use RAID-5. This isn’t because RAID 5 is particularly failure-prone, although it is less resilient than other RAID levels in some circumstances. RAID 5 arrays end up in our lab because they are very popular. RAID 5 is the most popular level of RAID because it is the most cost-effective RAID level to implement that provides both improvements in I/O speeds and fault tolerance.
This RAID setup uses XOR parity bits, spread across all of the drives in the array, to pick up the slack in case one hard drive fails. All of the parity bits use up one drive’s worth of space. Therefore, a RAID 5 array with six drives would only have the total capacity of five combined drives. RAID 5 has to perform XOR parity calculations all the time. However, these are simple calculations, and the time an effort needed to solve them are massively outweighed by the improvements in performance gained by striping the hard drives together.
RAID 5 has only one drive’s worth of fault tolerance, so if another hard drive fails before the first drive has been replaced (or fails while the first failed drive is being replaced), the entire array will crash. When our RAID data recovery technicians encounter crashed RAID 5 arrays, this is typically the reason for the crash. To recover your data, our engineers must determine how your array fits together and how the XOR parity data was distributed across the drives.
RAID 6 Server Recovery
RAID-6 a tougher and more resilient version of RAID 5. It is very fault-tolerant, as you are guaranteed protection against the failures of two drives. We see RAID 6 less often than RAID 5 in our RAID server data recovery lab. This isn’t just because RAID 6 is tougher than RAID 5, but also because RAID 6 is less cost-effective, so it is less frequently used. There are two layers of parity data, XOR parity and Reed-Solomon parity. These take up two drive’s worth of space. As a result, it takes more drives for a RAID 6 array to have the same capacity as RAID 5. RAID 6’s performance is also less improved, because it has to do many more parity computations than RAID 5. The second layer of Reed-Solomon parity relies on complicated Galois algebra, as opposed to simple XOR calculations.
When RAID 6 servers do show up in our lab, it is often because three or more drives have failed. This can be due to a power outage crashing multiple drives, or due to two drives failing while one failed drive is being replaced. RAID 6 is a bit of a tougher nut for our recovery experts to crack because of the two layers of parity data spread across the drives.
RAID 10 Server Recovery
RAID-10 is expensive to implement for your server, but provides immense performance gains. This nested RAID level is a combination of RAID 1 and RAID 0. The downsides of RAID 10 are its decreased capacity—half of its drives are used for mirroring—and the high cost of the drives needed to set it up. The upsides are massive performance increases—you get a read speed boost from RAID 0 and from RAID 1—and theoretically, the highest fault tolerance of any RAID level.
We say RAID 10 has theoretically high fault tolerance because in certain circumstances, it is less fault-tolerant than RAID 6. Due to mirroring, it is possible for a RAID 10 to keep going even after half of its hard drives have failed. But this is only if the right drives fail. In a RAID 10, each block of data is written twice, to two separate drives. If both drives containing, say, block #269 fail, the entire array fails. This makes RAID 10, in a worst-case-scenario, just as safe and reliable as RAID 5.
For RAID 10 recovery, our RAID server data recovery engineers need to deal with two RAID levels stacked on top of each other. RAID 10 can actually be made as a RAID 1+0 (a RAID 1 on top of a RAID 0) or a RAID 0+1 (vice versa). Either way, RAID 10 recovery requires our engineers to determine which parts of the array are mirrors and which are stripes, and how the drives are arranged.
RAID 50 Server Recovery
RAID 50 is much like RAID 10, in that it is also a nested RAID level with RAID 0. To imagine a RAID 50 array, take a RAID 0 in which several drives are striped together. Then replace each individual drive with a RAID 5 array made up of multiple drives. RAID 50 offers great performance, like RAID 10, because of the speed gains of RAID 0. It also offers performance boosts from each RAID 5 sub-array. RAID 50 gives your server more capacity than a RAID 10 array with the same amount of drives. Its theoretical, best-case fault tolerance is lower than RAID 10, since it can only lose one hard drive from each RAID 5 sub-array. Its worst-case fault tolerance, like RAID 10, is one hard drive.
RAID 50 is more difficult to piece together than RAID 10 because you have, essentially, stripes over stripes. Our data recovery experts have to puzzle together each RAID 5 sub-array while also dealing with the way the RAID 0 striping connects them together.
We generally recommend using RAID-6 over these nested RAID arrays. RAID 6 guarantees that your data is safe if two drives fail. It can also be more cost-effective to implement than RAID 10 and RAID 50. A RAID 6 array will have more space than a RAID 10 with the same amount of drives, and just as much storage capacity as a RAID 50 array with two RAID 5 sub-arrays. RAID 6’s only disadvantage compared to RAID 10 and RAID 50 is in its performance speeds. Nested RAID levels see the most use when your server requires the most performance and data redundancy is not as much of a concern.
Other RAID Levels
There are a few other levels of nested RAID that can be used in a server. Some of these levels include RAID 51 (a single RAID 5 array, mirrored like a RAID 1) and RAID 60 (like RAID 50, only the sub-arrays are RAID 6 arrays). Generally, these types of RAID servers rarely show up in our data recovery lab. This is not just because they are very fault tolerant (although they are). It is also because they are extremely expensive to implement compared to other levels. The cost of implementing these arrays is much higher than for other arrays, and come with dramatically reduced capacities and performance boons compared to their peers. As such, only a handful of people use these arrays for their servers. Users of these arrays tend to value fault tolerance above all else, even capacity and performance.
Windows Storage Spaces (Software RAID)
One new form of RAID array our data recovery engineers are beginning to see more and more frequently is Windows Storage Spaces. Storage Spaces is a software RAID. This means the array is set up and managed using software, as opposed to a physical RAID controller card. Storage Spaces offers massive advantages in flexibility over traditional RAID arrays. For one, the hard drives in your array don’t need to all have the same capacity. They don’t even need to have the same interfaces, or even be hard drives. You can theoretically make some pretty wacky (but useless) RAID arrays with Storage Spaces—like a RAID 5 made out of a USB flash drive, an external hard drive, and a 40-pin SCSI drive hooked up through an adapter cable.
Storage Space’s flexibility also allows users to make interesting RAID-like setups, not just standard and nested RAID levels. For example, you could have a setup with triple parity. You could have a single volume mirrored three or more ways. You could have a setup in which a portion of your Storage Space server mirrors data like a RAID 1 while another portion stripes it like a RAID 5.
The downside of Storage Spaces is that all this flexibility requires a lot of under-the-hood complexity. And that complexity can come back to haunt you if something goes wrong. Storage Spaces is a relatively new technology, which means our recovery engineers have had less time to develop our Storage Spaces recovery techniques. Over the past twelve years, we have encountered thousands of servers. Over the past four years, we have only encountered a handful of servers running Storage Spaces. That said, we are constantly refining our Storage Spaces recovery tools and techniques. The ingenuity of our data recovery engineers has yet to come up short.
RAID Server Data Recovery Scenarios
Regardless of which way you set up your RAID server, data loss is still possible. As long as hard drives can fail, no RAID array is immune to failure. You can have as many hot spares in your setup as you can fit, and something can still go wrong. Even if you exercise perfect prudence and are extremely lucky, and your server never crashes due to hard drive failure, somebody could still delete data from it. Or your SQL database or Exchange mail database could become corrupted. In these cases, you need RAID server data recovery services.
Gillware’s RAID server data recovery experts are world-class in their field. With our ISO-5 certified data recovery lab and SOC 2 compliant security policies, your hard drives and recovered data are as safe as can be. Our entire process is financially risk-free, and we even cover inbound shipping—no matter how many drives you need to send us. With our “no data, no charge” policy, you only pay us after we’ve successfully completed your case.
Servers are extremely important to any business, and extended downtime can be devastating if yours crashes. If you need your data back in a hurry, we offer emergency data recovery services. Our emergency services are just as financially risk-free as our standard services. In many emergency situations, we can get your most critical data back in your hands within a day of receiving your server in our facility.