I made this many years ago and just stumbled on it. Who remembers the early days of EBS volumes on AWS and needing to raid 8 of them together to get high throughput?
Is anyone still seriously using MDADM anymore?
what do you use then?
MDMA?
I use a CoW snapshotting filesystem instead - ZFS and BTRFS are two examples of this. Both support “raid” but not in the traditional sense
I was until about a year ago or so. I had all of my home server’s file systems on top of mdadm but somewhere around 3-4 years ago I started migrating my different storage groups to zfs and then I finally got rid of my root mdadm mirror last year, moving that to zfs too.
I’ve still got a 2 disk mirror that has been running for well over a decade. Started as 240gb I think and got upgraded in place to 4tb after a drive failure 5 or 6 years ago.
The server it’s on has been completely replaced multiple times and now mostly has zfs arrays but there has never been a justification to retire it.
It was the practical/easy way to make a software RAID in ubuntu 12.04, which is the era in which this was made.
Make Dat Ass Do Magic
