I never was much of a backup person until I started on the Vespers project. As things progressed, I realized how much content there was to manage, and how important it was going to be to protect it from loss. So somewhere along the way I became a backup fanatic. By my count, I believe I now have somewhere around six active backups of my data, some full backups, some partial. Let me see if I can remember them all.
My main backup is a clone of my desktop development machine. I use an awesome program called SuperDuper! for this. It basically just makes an exact clone of my computer’s internal drive onto a secondary internal drive. I have that running nightly, so at any point in time I have an exact clone ready to go, losing at most a day’s worth of work.
In addition to that, I also use Apple’s Time Machine software to create hourly backups of the main internal drive to a separate partition on the secondary internal drive. So if I somehow do lose a day’s work and it’s something I really need, I can access the Time Machine backup and get what I need. So in this case I would lose, at most, an hour’s worth of work. Not that I think I’ll really need a backup of that frequency, but that’s how Time Machine is set up, and that’s good to know.
Those are two pretty good backups to have, and that alone would constitute a decent backup system (except for the fact that they both backup to a secondary internal drive). For Vespers data specifically (and really, that’s the stuff I care about the most), I also create a few mirrors.
I use another nice program called ChronoSync to create scheduled nightly synchronizations of Vespers files to other drives. One copy is kept on my external FTP file server (the one my artists use for uploading and downloading), and another on a separate external drive I had hanging around looking for something to do. I also keep an up-to-date copy of the latest files on my remote MobileMe file space, so I’ll always have an off-site copy of things available in case of disaster.
Because I often work remotely on my laptop, it’s also nice to have a copy of the most up-to-date build readily available. So for this I use the excellent DropBox. That keeps a local copy of the latest build on my laptop, which is a mirror of the version on my desktop. So any time I make changes to the desktop version, it’s automatically updated on my laptop. It’s mirrored through a version stored on the DropBox servers, so there is also always an off-site copy available. I could easily do the same thing with my existing MobileMe storage, but DropBox is just exceptionally cool, and works a bit better (and faster) than MobileMe.
Of course, that’s kind of overkill. But once you get into backup solutions, it can get a little addicting. Once the home renovations are done and I’m back in the new office space, I’ll probably consolidate these a bit, but since they’re all managed automatically with scheduled sync’s, it’s not like it interferes with anything or is troublesome in any way. It’s good peace of mind.
The interesting thing comes when the backups are put to the test, as I am experiencing today.
When I woke up today, for some reason my main desktop machine was running, when it should have been asleep after the previous night’s backup routine. That’s never a good sign. It turns out there was an error during the cloning process, with a dialog box telling me that the main internal drive was full. I knew that there should be somewhere around 20-25 Gb of free space still available, so this was perplexing. After some poking around, it did appear that the main internal drive was being interpreted as completely full, for entirely unknown reasons. Where those available 20-25 Gb of space went is a complete mystery. And although the machine still appeared to be functioning normally, some weird issues started popping up — for instance, I couldn’t open my e-mail program since it could no longer find the designated “temp” folder. I ran some disk utility programs on the main internal drive, but no errors or abnormalities could be found.
I rebooted into the clone, and all appeared as it should. It still had the appropriate open space on the drive, and e-mail was working fine. So it would seem like the best option would be to revert to the clone. Who knows what other weird things will pop up with the main internal drive? And besides, I don’t know how I would go about finding where the missing space went and how to get it back.
Still, that would mean erasing the main internal drive and repopulating it from the clone. Cloning back from the clone, so to speak.
I’m not sure why, but that makes me a little uneasy. It shouldn’t—in fact, that’s entirely what the clone is there for. But I guess I’m just programmed to question what might happen when you lose the original and start working from a copy of a copy. It’s one thing to lose a file, or a folder of files, and to just recover a new set from a backup. It’s another to restore a complete system from a clone, or at least it seems that way. Still, it’s nice to have options.
Kind of interesting to think that my DirecTV receiver’s hard drive failed over the weekend, and now this. We’ll see if things really do happen in threes.
Enjoyed this article? Subscribe to The Monk's Brew RSS feed.
Wow, that trumps my three systems. My general rule of thumb (since losing the first version of Blue Lacuna back in 2006) has been that major projects are backed up on another computer in my house, a remote computer that I have full access rights too, and a private remove backup service like Mozy. The theory is that after any one point of failure (house burns down, Mozy goes out of business) I still have two other backups of the data.
The trick is to remember to verify every couple of weeks that all of them are still working. I’ve had all three silently stop working on me.
True, I think two local and one off-site backup is a good way to go. And great point about making sure all are continuously viable copies. I’ve gone a little overboard, but I’ll clean that up a bit in time.
Actually, you’ve got the right idea, but a mistake in implementation.
Dropbox isn’t a backup source. The data on there isn’t guaranteed at all. Since your primary backup is in the same PC, that’s a huge risk.
I’d suggest you drop Dropbox for backup and use something like Jungledisk or Mozy. Jungledisk is my preference since the Mozy software is kind of crap. It backs your data up onto Amazon S3.
Then for your primary backup, backup onto a different machine. Preferably one hidden in a closet somewhere, or if you want to get super fancy – in a fire safe.
That protects you from the two feasible ways you can lose your data – theft and/or PC destruction and theft and/or destruction of your primary backup. Remember, hard drives are fragile – and if you have both in one machine, all it takes is one power supply going crazy to lose both of them.
@Jason: Excellent comments. I don’t actually use Dropbox as a backup source, I was just including the copies of files stored with Dropbox as a potential source for files if any of the backups failed. Because of Dropbox, I have three sets of Vespers files available (one on the desktop, one on the laptop, and one with Dropbox) if any/all of my main backups happens to fail. But you’re right, it’s not a true backup source solution. I use it in order to keep an up-to-date version of the latest Vespers build (and a number of other unrelated files) available on my laptop.
I also agree that it’s not a good idea to group all backups within the same machine, and this is why I keep two backups on external drives. One of those is actually a NAS device, which I keep separate from the desktop.
I also put one backup on MobileMe, for the off-site copy. I don’t particularly like MobileMe, though, so I’ll look into Jungledisk.
Ya my bad, on a re-read I see that. 🙂 I personally don’t trust any backup that isn’t redundant myself. For instance, I used to think (long ago) that uploading to my server was fine.. Until I had several RAID-5 clusters die. Irrecoverably.
Now I’m paranoid about it like you. 🙂