the real question
Wil itl be truly production ready before btrfs? I know different use case but couldn't resist making fun.
The HPC-focussed parallel cluster file system, BeeGFS, has received certification to run over Intel's OmniPath Architecture (OFA). The cert, plus a collaboration with Intel's OPA team, is a step forward for BeeGFS, which was spun out of Fraunhofer in 2014 and made the responsibility of ThinkParQ. BeeGFS runs on Linux distros …
asdf, I get the joke.
BeeGFS (and other parallel filesystems such as Lustre) are at a layer above the filesystem.
The storage servers in the documentation are recommended to be RAID6 with an ext4 or xfs filesystem. Or could be ZFS.
I don't see why btrfs wouldn't work - and would be interesting when it is production ready.
Having not been following it closely lately Iast I heard the big problem with btrfs actually is with RAID. Also from what I understand it is fairly stable for most setups but what worries people the most is when it fails it often fails hard core and often makes data retrieval virtually impossible. Not a feature you want in an enterprise filesystem.