Goodbye to RAID?

We built a cloud infrastructure for our cybersecurity teaching, and it has scaled up well. Overall, we learnt so much in doing this, and one main lesson was the RAID (redundant array of inexpensive disks) 10 was your friend. Why? Because disks will fail, and RAID 10 integrates error correction across a number of disks, and when one disk fails in an array, the data storage infrastructure keeps working, and there isn’t too much of a performance hit. Along with this, servers fail, too.

To overcome server failures, we used a cluster approach, and where there processes do not depend on a specific server. A failure of a server just causes the total compute power to drop, but, if the loads are manageable, there should not be a drop in the overall performance of the cluster. Often, if you were lucky, a server would last between 3–5 years and then be replaced.

But, using RAID and replacing perfectly good servers has a carbon footprint and affects e-waste. Now, a Cloud service provider — the Paris-based Scaleway — has defined that they intend to remove RAID controllers and keep their servers for 10 years. Overall, Scaleway has over 14,000 servers, so the savings in waste could be significant: