CRFS and POHMELFS
CRFS and POHMELFS
Posted Feb 7, 2008 7:20 UTC (Thu) by heini (guest, #33614)Parent article: CRFS and POHMELFS
Looks like both projects try to reinvent AFS.
Posted Feb 7, 2008 9:41 UTC (Thu)
by nix (subscriber, #2304)
[Link] (3 responses)
Posted Feb 8, 2008 3:17 UTC (Fri)
by linuxbox (guest, #6928)
[Link] (2 responses)
Posted Feb 8, 2008 13:03 UTC (Fri)
by Velmont (guest, #46433)
[Link]
Posted Feb 9, 2008 22:42 UTC (Sat)
by giraffedata (guest, #1954)
[Link]
Posted Feb 7, 2008 10:41 UTC (Thu)
by IkeTo (subscriber, #2122)
[Link]
CRFS and POHMELFS
AFS, with a much more efficient protocol, not a horror to administer, and
actually making an effort to be a POSIX filesystem rather than
gratuitously reinventing things like, oh, permissions?
Seems like a good thing to me, although you could replace `AFS' with 'a
distributed filesystem' and get the same answer :)
CRFS and POHMELFS/AFS
A more interesting comparison for me would be CRFS vs. GFS or OCFS--which, I rather suspect,
is more what the btrfs authors are aiming at.
CRFS and POHMELFS/AFS (what about HAMMER?)
How does HAMMER fit into all of this? Yes, it is being developed for DragonflyBSD, but it
could maybe come into Linux. Does anyone know? :-)
I'd like to know how it's better than NFSv4. The only reason NFSv4 exists is to solve those classic NFS problems. It definitely has client-side caching and POSIX inter-user synchronization of file access.
CRFS and POHMELFS/AFS
CRFS and POHMELFS
From AFS FAQ:
> Subject: 2.01 What are the differences between AFS and a unix filesystem?
> ...
> Authentication: [ User ]
> ...
> File permissions: [ User ]
> ...
> Data protection with AFS ACLs: [ User ]
> ...
> Protection groups: [ User ]
> ...
> Hard links: [ User ]
> ...
> Changing file protection by moving a file: [ User ]
> ...
> chown and chgrp: [ User ]
> ...
> Save on close: [ Programmer ]
> ...
> byte-range file locking: [ Programmer ]
> ...
> whole file locking: [ Programmer ]
> ...
> character and block special files: [ SysAdmin ]
> ...
> AFS version of fsck: [ SysAdmin ]
> ...
Is this the type of things that CRFS explicitly says they want to avoid?