I haven't used rdiff-backup, but I used to use rsnapshot (actually a homebrew equivalent to it) -- was backing up several hosts to a central one. But I really missed having all the backup metadata in a database, where I could do simple SQL queries to find out file patterns were taking up the most space (this helps you tune your include/exclude list). Also, trying to replicate a rsnapshot volume that had a bunch of hard links (each day's backup's common files were hard linked to the previous days' files) -- this made for some very slow copying, unless I did a raw image copy (30 systems, with 10 daily, 6 weekly, and 12 monthly backup each made for a lot of file inode entries). That's why I wrote Snebu, so for each file that doesn't change between backups, only one gets stored. And references between backup sets are handled in the DB (sqlite3 based) instead of via hard links in the filesystem. Oh, and files are also compressed (lzop compatible format), which is something that rsnapshot didn't give me.
My favorite feature, that I'm testing out now (should be in the next version once it is stable and I hammer out the UI issues) is the ability to have a shadow copy of the backup DB that you stick on a thumb drive. This allows you to make incremental backups of your laptop to the shadow copy and sync it back to the main backup later on. Other features coming include external plugin modules to support moving / copying older backup sets to independent volumes, and potentially tape changers and cloud storage too (however these will all be secondary storage locations, the primary will be local storage).