Make sure that your ContentCleaner is working and clean out the contentstore.deleted directory Make multiple content stores and keep most of the content on a slower storage mechanism Write an archive-aware ContentStore implementation
The contentstore directory structure is ordered by date. Alfresco can delete files but will never overwrite files i.e. every content update actually creates a new file and the old one is left to be cleaned up later: http://wiki.alfresco.com/wiki/Content_Store_Configuration
So, you can concentrate your backup to a more specific directory, safe in the knowledge that you have backups of dates prior to that, and that Alfresco will not write new files into the earlier-dated directories.