Linux backups 
I've been working on writing some
backup scripts in
Python for the product at work and came across
this really neat article which talks about how to do
incremental backups on Linux. The central idea is the fact that
filesystems on *nix are similar to garbage collection in Java in the sense that files are stored somewhere and filenames are just pointers to the actual data. When there are no more pointers ( filenames or
symlinks) which point to a file, it is automatically removed from the filesystem. This leads to some really interesting side effects as evidenced by this article in terms of incremental backups - very cool!
http://www.mikerubel.org/computers/rsync_snapshots/