The issue / use-case is distributed filesystems.
These do NOT need ANY interaction with the system, the system does not use this space.
Most programmers, if not programming a system tool, will use the OS standard calls for ANY file system operation (eg copying, moving, creating, opening, editing, deleting a file). These are always the same, no matter which file system is used.
So: no matter if ext4, btrfs, xfs, zfs is used, most operations will be standard system calls, the same for ALL filesystems…
At the moment, ALL so called “standard” file systems are NOT “distributed”.
If a SME company has 2-3 sites, like I have a few of those type of clients.
With NethServer, the main site will have direct access for eg Windows Clients.The other two sites would have to access via (very slow) WAN connections over VPN.
An iBay using MooseFS, mounted as usual under /var/lib/nethserver/ibay/moosefs, could be synched to a Raspberry PI, a NAS, another NethServer, and users at those sites could also have fast, direct access.
The sync (Delta-Sync) would be from MooseFS to MooseFS, one of them running as “master”.
Nextcloud would allow client-sync, but with 10 clients at one site, it’s MUCH better if the there is only one sync in the background for ALL users at site B. NextCloud would have 10 different sync operations ongoing… (Danger of Race-Condition).
Also good to note:
Most Devs do not go into system operations. These are “reserved” for Kernel-Coders or Tweakers…
Here on NethServer forum, I’d call @Stephdl a top coder. However, I may be mistaken, but I do not think he’s ever dabbled into filesystem coding…
My 2 cents
Andy