This has been on my mind for a while. I need to implement an efficient, and easily scalable, backup solution. My data is just too valuable to not be backed up. The only thing is that I haven't found THE perfect tool yet. Here is my dilemma:
i) of course, backup should be automatic. Oh, yeah, versioned and incremental. In fact, I want to have a choice of which parts of it should be versioned, and which ones are simple copies of the original data. For example, I probably want my programming and research data to be versioned, but I don't need versioned backups of my music or pictures.
ii) it should be easily replicated into multiple locations. Ideally, I plan to have a server at home, and put another one on my dad's house. Maybe we could exchange backups, I'll backup to his machine, and let him backup to mine.
iii) easily retrievable, and "platform independent". I run on a Mac right now (two, actually). But my dad runs windows, as does my wife. Nevertheless, I still use Linux pretty much every day via the network, and also Windows, and Solaris. I don't want my data to be constrained in a proprietary format that needs one specific program to restore a backup. Worse, if that program runs in a mac, and my mac is gone, I'm screwed.
iv) it should be able to retrieve and send data via the network. Ideally, over SSH. I hate having to configure a SAMBA server, mount the disk, do the backup, and unmount it after that. If I don't have to mount it, it seems to me it would be a little safer against accidental deletions.
v) the cheaper, the better.
All right, all the paths lead to rsync. This is what I've been using lately. Apparently, it works (yeah, it has some bugs on the Tiger version, but I'll have to live with it). I wrote a bash script to backup my main data folder (no pics/music are included, that part is still manual). The problem is, right now only my main folder is backed up. I've got data floating all over a number of machines, and those are not backed up (at least not by me, as I'd like to think that the sysadmins do it pretty regularly)
My current script is not very flexible. And bash is not exactly the best scripting tool for elaborate scripts, I think (well, maybe I'm not that fluent in bash, hehe). Therefore, I've been looking at other solutions.
Here is a good assessment of the tools available for the Mac. My problem is that I do not want a solution that is Mac-based. It has to run on multiple platform.
So far, I like rsyncbackup. I plan to deploy a solution based on it.
Now, the last problem standing: how do I synchronize my multiple computers? I wanted to have the same data pretty much everywhere. I still want to find the best solution for synchronizing my computer (don't suggest me .Mac). It is specially difficult if I want to propagate deletions.
UPDATE: Hm… I think I found a good synchronization tool!