Local Cloud

In Step 2, I described how my rsync backup script functions. Why do I configure my rsync backup this way? Well, as I may have already described in more detail, I was a Mac user for years before switching to Linux. Before I switched to Mac OS X, I used a very haphzard backup method. For those times I bothered to have a backup, I would copy my data manually to a floppy, SuperDisk, Jaz drive, or external hard drive. Tedious, slow (if you want to bulk copy, Mac OS did not merge folders back then, just replace everything), and was not automated.

Once I switched over to Mac OS X, I initially used an excellent tool called SuperDuper! to automate my backups. The paid version offered incremental and scheduled backups. For a time, this setup worked well to a locally attached external drive. As SuperDuper! did not offer a network feature implicitly, I was able to turn on network sharing and mount a sparse image for backup when I decided to advance to network sharing with storage attached to a spare Mac (first a Power Mac 7300 and then a Power Mac 7600 if I remember correctly, both G3 upgraded). Enter Leopard and not every app was ready for the transition at launch, cue transition…

Having used Carbon Copy Cloner for a client, I was familiar with that tool as well. After mostly on account I copied the structure of how Carbon Copy Cloner worked when I used a Mac. It created a folder for the backup and then a series of date and time stamped folders for each archive of deleted/changed files. Honestly, I probaly should have simply went the hard link route with each dated folder representing a snapshot of my home directory at that moment. While each snapshot would appear to contain many duplicate files, only changed/deleted files will have been written each time, with the rest of the files/directories exsisting as hard links to other points on the backup drive. Many pointers, only one copy of the data.

As such, let me discuss how that model would work

Section Headline