Erasure coding vs. RAID: Pros and cons for hyper-convergence

Erasure coding vs. RAID for HCI: Capabilities

Consider the system's capabilities first. Not every hyper-converged system supports erasure coding. Although erasure coding is becoming far more common, it's not yet universally supported. Assuming your hyper-converged system supports both RAID and erasure coding, the next step is to figure out which is a better option for your platform.

RAID is a mature technology that has been in use for decades. The problem with using RAID, however, is that depending on how it's configured, it may have extremely high overhead requirements or offer limited protection. A RAID 5 array, for example, can protect a hyper-converged system against the failure of a single disk. If a disk fails, though, the amount of time required to rebuild the array once the bad disk has been replaced can be excessive. During that time, the system is vulnerable to the failure of an additional disk.

Conversely, a RAID 10 array can protect against multiple simultaneous failures, but has 50% overhead. If such an array were composed of 10 disks, for instance, then five of those disks would be used solely for mirroring data. That's a lot of wasted capacity.

Erasure coding, on the other hand, allows you to define the level of protection you want to put in place for your data. You're not limited to using rigid data protection schemes such as RAID 5 or RAID 10. Additionally, erasure coding tends to make far better use of the available space than a RAID solution based on mirroring would.

Erasure coding vs. RAID for HCI: Configuration

After considering these pros and cons, it is necessary to consider the way your hyper-converged system is configured before choosing between erasure coding vs. RAID for data protection.

RAID structures are best suited for the protection of an individual storage array. Erasure coding, on the other hand, is a much better fit for scale-out systems like hyper-converged infrastructure. This is because erasure coding can spread redundant data across available nodes.

It's worth noting that erasure coding tends to be a bit more CPU intensive than RAID. So erasure coding also may deliver a slower write performance than RAID data protection.