S3. I have a script that takes one argument, a directory name.
It encrypts the directory name and adds a timestamp, to make a filename that reveals nothing about the contents of the file except for the time it was created. It stores the original directory name along with the encrypted directory name in a local log file, so I can tell what's what later.
Then it zips the directory and its contents into a file, names it with the encrypted filename, and does a GPG encryption of the file.
Then it uploads it to Amazon S3 and cleans up after itself.
All of this is completely automated once I type in the command. I do it for my project directory, which contains all my projects for the calendar year, periodically. After the year closes out, I start a new year and the old files (from previous years) stay safe and unchanged on S3.
Over the course of a year, if a few backups for the same year pile up, I usually delete the old ones. I haven't automated that part.
This scheme is OK for backing up a few important directories, if they aren't large, like for source code, just what you asked about -- a few hundred megabytes will cost just pennies a month on Amazon.
It's not a SVN or GIT solution, but it gives you total privacy, rather than relying on a third party to do the encryption for you.