Data Deduplication File Systems

[Michael Torrie]

On 02/24/2013 11:45 PM, DANIEL DAVID EGLI wrote:
> It only dealt with one file system, called lessfs, which (at least 
> the way they showed it) could not be mounted via fstab. You had to 
> run a separate binary to mount it. Has anyone heard of any other file
> systems that implement data deuplication?

Not addressing your question directly, but what the article described
was a "fuse"-based filesystem.  The entire filesystem runs as a program
in userspace using the fuse framework (yum install fuse or aptitude
install fuse), instead of in the kernel.  A lot of popular filesystems
use this these days including sshfs including ntfs-3g.  And actually you
can indeed put them in fstab, if you make sure there's an appropriate
/sbin/mount.<fsname> script or program.  I use sshfs quite a bit, though
I always invoke it from the command-line because of its nature
(accessing arbitrary remote servers on an irregular basis).

I've often wished that there were fuse versions of every filesystem in
development, so I could easily test them on older distros without having
to compile new kernels.  Like for btrfs.  It would also allow file
systems to be ported to OS X and maybe even Windows.  I know performance
would not be as good.  I think it would still be an interesting
development tool.