RAID 5 Rebuilds
An important fact of life is rebuilding degraded RAID arrays. When a member disk fails or needs replaced, doing a rebuild task can be very time consuming. Here we timed a rebuild of the RAID 5 array with 8 x 256GB C300 SSDs.
As illustrated here, the 1882 shows a roughly 23 percent improvement in rebuild speed. This is with only 20 GB of data on the array, as we are just trying to illuminate a percentage of improvement. For extremely large volumes, when rebuilds can stretch into time-frames that are measured in days, not minutes, an improvement off this magnitude is very advantageous for the end user.
DEGRADED RAID 5 PERFORMANCE
Another important aspect of the 1882 is the fact that it performs very admirably when in a degraded state.
For those not familiar with a degraded state, it is simple to cause the controller to enter a degraded state.
You simply pull out one of your drives! While we were running a benchmark I simply pulled one of the C300s out of the enclosure. Te array will continue to work in this state, as that is the purpose of RAID 5. Data protection is what it is all about, and being able to survive a loss of a member drive and continue function. However, the controller will have to make intense calculations constantly to calculate the parity data, so that the array continues to work fine.
In a degraded state, the controller will perform slower, as there is a ton of parity overhead calculation being dropped on the ROC. This is where strong controllers shine, still delivering a fast service to the user. It is also where weaker controllers will fail, becoming terribly slower and less productive.
The 1882 certainly raises to the challenge in this testing scenario!
Below are some benchmarks that we conducted with a degraded RAID 5 array. The 1882 performs very well in this category. Clicking the results below will allow you to scroll through the results.