I’m currently using a few different filesystems for different use-cases, and looking forward to bcachefs because (as I understand it) it covers some use-cases that I’ve been dreaming of for ages and not yet found any solution for:

- raid5/6 with odd-sized disks (btrfs does raid1 with odd-sized disks, mdadm & zfs do raid5/6 with equally-sized disks) - great for a homelab server that I upgrade ship-of-theseus style rather than the enterprise “buy a whole new server with whole new larger disks in one go”

- SSD as a cache integrated with HDD RAID (right now I’m using bcache + mdadm + ext4 for this usecase; having the filesystem coordinate all three of those layers feels more elegant and efficient)

- Different settings for different directories in the same filesystem (my super-important documents get RAID1 across 4 drives, my blu-ray rips that I can just re-rip if I lose them get RAID0, everything else gets RAID5)

It's got about the same features as btrfs, and other new-gen filesystems like ZFS, APFS and ReFS, such as:

- Copy-on-write

- RAID 0/1/5/6 equivalents

- Subvolumes

- Snapshots

- Transparent compression

- Reflinks

- Metadata and data checksuming

- Replication

- Bitrot protection (when used with RAID and checksuming)

And it's supposed to be more though through and well designed than btrfs.