Today I got to continue my quest for perfect backup. Today, I'm taking a look at duplicity. I also wanted to further investigate dibs, which I remember from a while back. Even though it has an interesting idea of backing up to a data pool in the internet, it doesn't fit my goals so I'll put more effort on duplicity and add it to my [[!list of features in backup solutions|2007/backup_comparison]].
Emacsen commented my previous [[!backup post|20070611_backup]]. He named 3 scenarios for backups: Human error, hardware failure and archival, and focused on the first 2. Even though snapshots are useful for many purposes, I wouldn't use them for backups. Human error is really the only case where they come in handy (when one considers using them as persistent snapshots). Using snapshots as a tool to speed up backups instead is a good idea.
In my case, I'm only preparing for hardware failure. Past experience has taught me that hardware will eventually fail. And with the increasing disk sizes the amount of data lost on hard disk failure keeps increasing too. I've had my share of hardware failures that caused me to loose data (directly or indirectly), so I'm focusing on it.
In the end, I don't make mistakes that make me loose data too often. I've learned from my past mistakes and use VCS (and now DVCS) when possible and sort my files properly to reduce the risk of making mistakes and running in to those "Oops!" situations.