Can you back this up with some sort of data or a link? This sounds like pure speculation. Some MTBF rates or any other empirical data showing the Drobo failure rates. Not saying that they do not fail, but you make a bold claim without backing it up.
It's not a bold claim, it's elementary mathematics: A Drobo contains 2 drives and 1 set of computer logic, power supply, etc. A Drobo (or any other RAID solution with redundancy) protects only against hard drive failure (which is a great thing!). If the Drobo's power supply dies or an electric jolt destroys the logic board, the Drobo is dead. Hence, you have a single point of failure.
So for example if different components die independently of each other, the probability of that happening at the same time is very low. If, for instance, you don't have a surge protector and a surge strikes both, your hard drive and your computer, both may die at the same time. That's because the two events are strongly correlated. In case of my parents, a faulty USB hub killed their Mac mini and because the machine was apparently just doing a backup, also the Time Machine backup volume. Fortunately, the main drive was intact and I was able to restore everything.
I keep one backup drive at work and one at home. Even if a lightning strike completely fries my electronics at home, I still have a backup at work. Also in terms of components, it's safer: unlike the Drobo, I have
two controllers plugged into different electrical systems. If one controller fails, I still have a second one in a different enclosure.
RAID or RAID-like solutions serve two main purposes:
(1) You can keep on working in case a hard drive fails. This is not important if you use the drive for backups as you never work off backups. To some people, it may be worth it to mirror the boot drive, for instance.
(2) It provides more storage capacity in case you need a volume that is larger than the largest hard drives available at the time. If you need a 4 TB volume, for instance, you need to combine several drives.
Then, redundancy becomes crucial, because otherwise one dead drive will kill the whole volume -- and the probability of failure grows roughly linearly. If the rate of failure for a single drive is, say, 5 %, then to leading order, the probability of failure grows linearly with the number of drives. To counteract that, you introduce redundancy. That's where I see the Drobo.
Just to be clear here: I like the Drobo, I think it's a great concept and a great product. If I had the money, I'd probably get one. But in this specific situation, I think the OP fares better with a different solution.