Currently we backup our user's data via cross server backups... IE:
ServerA is backed up on ServerB
ServerB is backed up on ServerA
Recently we've found that certain nodes are much heavier backup wise than others and so we decided to get a backup node, but we realize that of the 36 TB we currently have floating around on our nodes a lot of that is probably torrents and the operating systems themselves. Which leads me to my question: What, if any filtering do you preform on the user's data before backing it up?
I'd like to apply a filter to the files backed up so that:
- Files over 250 MB are skipped
- The user's base OS is removed from the backup
- Any folders labeled "torrent" and possibly a few other names are scanned for manual review by administrators.
I've also thought about stating "no backups, you are responsible for your files to either your own solution or to our backup location" and offering a 20 GB backup space per VPS along with a example script for automated backups.
Does this match what you do or is this far too strict for backup purposes?
Just as a final note, we haven't decided on any one thing yet, so any of our customers are also welcome to give us feedback.