The problem with a differential backup is that you are either screwed sooner or you are screwed later
If you don't keep *every* backup job, including the original full backup, you can't ever restore to a clean state. So you are either unable to restore or unable to backup (due to disk space.) Of course you could keep buying more storage, but then you may look up one day and to restore something you need an entire datacenter to server the backuped files
Have you considered Subversion or CVS? Currently I'm using Subversion to backup my development projects, web site source, home directory and documents, and business documents. It works pretty niceley. Subversion has some pretty slick ways of dealing with copied and modified files, so storage space isn't so bad.
For awhile I had an automated backup script, but I axed it. I'm going to re-instate it for certian parts of my backup set (home and business directories). The script wen to each directory, added changes to the archive, then shut down the archive (I host my archives so I can grab them remotely if I like without using SSH) and copy + compress them off.
Archived topic from Iceteks, old topic ID:3753, old post ID:30562