I've been running a 9TB RAID (the old apple promise version) since 2007. Had my first drive failure this year. It was a hassle, but didn't lose any data. However my backup is not "live" so I was basically not working for several days while the drive re-built itself followed by a bunch of diagnostics. I've since developed a new solution to the problem of working during a failure/repair cycle.
1. The current 9TB raid is now my working drive- it is basically I/O and temporary storage setup as RAID 50
2. I have a second RAID that is a 12TB thunderbolt device that is used to mirror the 9TB working drive
3. The Thunderbolt is managed by a Mac Mini using XSan and connected to the fiber channel raid via a FC slikworm switch
4. The thunderbolt RAID is backed up to LTO
I still edit with the Mac pro and the mini backs up my projects. As I finish a project it gets archived to LTO and the client gets a copy on portable hard drive. Now if I have a failure I still have live data that allows me to continue working. Instead of rebuilding the drive I can wipe the set and restore the project from backup. Alternatively if something happens with the backup, I'm still protected by the RAID 50 config which will allow me to rebuild from parity.
Also- don't go DROBO. Probably OK for a stills workflow, but not motion.