I've been trying to go through this thread and didn't see a description of what the individual drive failure entailed. What exactly happened to the 'failed' drive when it the failure was detected? Does it no power up anymore? Doesn't spin? After you pulled the 'failed' drive from the DROBO, did you try checking the disk in a standalone computer using the manufacturer disk utilities to run SMART checks on it?
It's a lot easier to troubleshoot where the problem is occurring if you can isolate what the resulting failure is. If the drive 'fails' in the DROBO and you yank it, run checks on it, low level it, and there are no bad blocks or SMART errors thrown, and put it back in the DROBO and it fails again then you know it's not the drive.
Then you can begin determining if it's the DROBO itself, or cabling.
As far as RAID, RAID is an excellent method for insuring data retention. It's not fool proof, but if you plan for contingencies, RAID is useful. The people who remark that you can't access data on the disks in a RAID5 (or similar design such as the DROBO), yes that's true. But you're getting N-1 disk utilization vs N/2 in a mirrored RAID system. You have to balance cost-benefit ratio to see if RAID5 will work for you.
For pure redundancy, RAID1 is the way to go. It's simple and it works. You lose a bit of I/O performance in writes, and you gain nothing in reads, but it works.
But in *ANY* RAID setup, you *NEED* to have a hot-spare hard drive laying around. If a drive in a RAID failure setup goes down, just because it's a mirrored stripe and the other one is working, doesn't mean you should wait a day for a new drive to come in. You should replace the failed drive immediately.