Physical drives fail, it’s a fact of life. It doesn’t matter whether it is spinning disks (HDDs) or memory chips (SSDs) that power your server hardware, drive failure is a high risk and you simply won’t get warned.
Yes, drive failure is less of a risk than it used to be, but the impact of drive failure is no less than it was in the past. Thankfully there is an easy fix in the shape of RAID, albeit a fix that can be costly in terms of storage space.
What is a RAID drive?
RAID combines two or more physical drives into one logical RAID drive. RAID is used to both improve disk performance and to protect against disk failure.
Drives rarely fail simultaneously but it is not uncommon for a single drive in an array to fail. Unfortunately for most workloads the implications of a single drive failing are just about equal to the implication of all drives failing – your server goes down.
RAID (redundant array of independent disks) relies on the fact that physical drives do not fail simultaneously. By redundantly storing your data in exchange for sacrificing some storage capacity RAID can protect you against individual drive failures.
How RAID Works
There are a number of different RAID configurations you can pick from, but all RAID configurations have a common element: RAID stores the same data in multiple places to ensure that the failure of a single drive does not sacrifice the integrity of your data. These are the important characteristics of RAID arrays:
Multiple physical drives: A RAID array contains multiple drives by definition. Though a RAID array can contain as few as two physical drives most arrays contain more physical drives to offer a better blend of performance while sacrificing less storage space in achieving redundancy objectives.
Firmware controller: Though software, OS-driven raid is entirely possible most RAID configurations will use a hardware controller that manages the RAID array. Drives are attached to a configurable controller which manages the array, redundantly storing data and bypassing failed drives to ensure continued operations.
Resistant to drive failure: Except for RAID 0, if a drive fails the RAID controller notifies you of a failure. You then have the opportunity to replace the drive after which the RAID controller will re-build the array, restoring it to full working condition without data loss.
Transparency: From the user and operating system perspective a RAID array simply appears like a standard volume. The underlying complexities of the array are invisible, your RAID controller presents an ordinary volume to the operating system while managing the redundancy aspects of the array behind the scenes.
Important RAID definitions
Let’s look at a couple of RAID definitions before we discuss the different RAID levels in further detail:
Mirroring: At a basic level mirroring involves the exact duplication of data on more than one physical drive to ensure that the failure of one drive does not lead to the failure of the array. Mirroring ensures data is available in more than one place.
Striping: RAID striping involves dividing data into blocks that are in turn spread across several disks. Striping improves performance because it allows multiple drives to simultaneously respond to a request, taking advantage of the combined throughput of multiple drives.
Parity: RAID parity calculations are used to ensure that data can be fully retrieved from your array even if there is a physical disk failure. Parity data is not an exact copy of data and therefore differs from mirroring. But RAID parity information is sufficient for the reconstruction of data in case of drive failure.
Nested RAID: Hybrid or nested RAID levels combine the features of more than one RAID level in an attempt to combine the key advantages of both levels. RAID 10 and RAID 50 are examples of nested RAID.
Explaining the different RAID levels
RAID levels refer to the different ways to configure a RAID array. Higher RAID levels are not necessarily better, each level has different advantages in terms of performance and redundancy. There is no right or wrong RAID level, it depends on your use case.
