I have a xfs filesystem with many small and a couple very large files (~100GB), its on a lvm2 volume. I want to do incremental backups from it. Is there any computational overhead when using a tool like for example duplicity/rdiff versus xfsdump for the backups ? In short will xfsdump reuse some fs internal information to significantly reduce the work it has to do to get the deltas or can I as well use something with nice features without significant performance penalties ?
