Without addressing the drobo, since I have zero experience with it, I will say that not backing up your data is a serious error, even with a perfect RAID implementation. (A point of reference: my day job is building a RAID NAS device, one of the big ones like what Jon mentioned, where loss of data is not realistic.) That's because RAID protects you against one kind of problem, but there are many other types. The most obvious example is that the RAID can protect you from a disk drive failure (or possibly two, if you're paranoid and willing to invest). But if you have a different type of problem, for example you delete the wrong file, the RAID system obediently deletes what amount to all of the copies - as quickly as possible. And you have no method of recovery. Other forms of this class of error are an OS or application bug, etc.
The way the mission critical businesses do things can be scaled down to folks like us:
- first, have your primary copy protected by RAID. That could be a mirror (RAID-1) or something more complex.
- next, make copies of your data occasionally. Ideally those are "snapshots" or "point-in-time copies" provided by your RAID or your operating system, but they could be DVDs or even other hard disks. This is crucial to protecting against the errors noted above.
- If you've done this much, you are still subject to certain types of disasters, for example if there's a fire in your office, chances are pretty good that either the heat or the firefighting methods will destroy all of the HDD, RAID and very possibly all of the copies.
- To get around this, the big businesses replicate their RAID volumes to another location, a "considerable" distance away. The NYC banks, for example, mostly replicate further away than NYC and often much further - one that I have worked with extensively has a backup data center in Nebraska, and others go across the Atlantic.
- Fortunately they also had off-site copies, meaning in that particular case magnetic tape stored literally under a mountain many miles away. We can simulate this by having a couple of HDD that get a copy, say, every week from the main RAID system, and then they are taken home or to the safe deposit box at the bank, etc. With two of them you can always have one at the bank even while you're loading up the other one.
So there you have it - four HDD and some software. Two in a mirror on your primary system. An offsite copy, possibly one that includes periodic snapshots, and another one that is in rotation with the offsite.
As far as your current predicament, there are services that specialize in recovering things like this. If Drobo themselves do not offer something like this (if it's their bug they ought to do something for you), there are plenty of others who do. They can usually recover stuff even from disk platters that are physically damaged, although there's a limit to that. (And there's a price, too.) I've seen data recovered from a disk that had a completely dead embedded controller (meaning the circuits glued to the bottom of the disk drive) - it did not even spin up. When they had the platters grafted onto a substitute controller, they discovered that there had also been a "head surface interaction" - what we more descriptively call "a head crash" in which the read/write heads physically land on and damage the recording surface on the platters. These folks were desperate, so they had their service recover all of the data that was actually on the HDD surfaces. They got something like 98% back, the lost parts being what was under the head crash. Clearly you should not have to go this far.
My guess is that if you haven't done too much to the remaining components, it's probably fairly straightforward for some expert to go in and liberate most of the data. I'd guess that it would cost $500-$2000 depending on just what went wrong. Quite a bit for an amateur and not really so bad for, say, a professional photographer.
_____ Brian... a bicoastal Nikonian and Team Member
My gallery is online. Comments and critique welcomed any time!