Drobo bills itself as a DAS device that automatically protects your data in case of a drive failure. Last year, I purchased a first generation Drobo, and over time filled it with 4 Seagate drives, all purchased from the Drobo Store, 3 750 GB drives and 1 TB drive. Last month I was copying about 100 megs of data onto the drive, and the copy froze and the Drobo rebooted itself. After it came back on, the drive was corrupted and no longer accessible. I ran chkdsk, and it went through its procedure of cleaning up a lot of bad sectors. After it finished, the drive was now accessible, but only about 11 GB of data was there, when originally I had over 1.6 TB. The blue capacity lights still showed that it had all of my data, but I wasn’t able to access any of it. I scratched my head, filled out a customer service ticket and went to bed. The next morning, I woke up, and the blue lights were gone. No data, no evidence of data, nothing. I went back and forth with customer service this past month. I sent diagnostics, screenshots of the drive lights, and event logs from my system tools. In the end, they were unable to determine what went wrong. As such, they decided the error was not with the Drobo and therefore I was on my own.
WTF!!! THE WHOLE REASON I BOUGHT A FUCKING DROBO WAS TO FUCKING PROTECT MY DATA. After getting into it with several people on the drobo forum, their answer was basically “it’s your fault for not backing up your drobo.” The whole point of buying a drobo was to back up my data! If I had to run a separate backup utility, I never would have bought a drobo in the first place!!!
Worthless product. Worse than worthless, dangerous. I had external drives for 7 years and never had a problem. I had a drobo for 1 and lost everything. PLEASE do not buy this product. Get a NAS or software raid solution, or just run redundant externals instead. I wish I had.
Uh, RAID definitely is no substitute for backups. You’re better off buying two external disks, backing up to one regularly (say, daily), and then backing up to the other one on a less frequent basis (say, weekly.)
Hell, I can’t even tell from the Drobo page what RAID levels it uses internally. Caveat emptor, man.
See this is what I don’t understand: If raid is no substitute for backups, WHAT GOOD IS IT? Why would I want a raid? If I am doing backups, then I can handle an external crashing, so what does the raid do?
I confess to ignorance on how this thing works—is it not fully redundant?
I assume it uses some combination of striping and mirroring… what generally happens to a striped/mirrored array when there’s a disk failure? Did you have multiple disks fail?
I haven’t actually found the exact mechanism they are using - they claim it is a proprietary algorithm. To be perfectly honest, I don’t know exactly what happened. My only guess is that the 1 TB seagate failed in a weird way that the drobo wasn’t equipped to handle. I know there is an issue with seagates and drobos (which is weird, because drobo sells them through their store, which is part of the reason I went with the drives), and the 1TB seagates in particular have had their own issues, but after a month of back and forth with drobo, nobody has any idea what exactly happened to kill the data.
RAID is good because (in most configurations) you can swap out a broken drive and restart the machine without any interruptions - and some setups allow hot-swap as well, so you don’t even need to shut down your computer.
Also, it can increase read and/or write speeds and increase the total volume (make 4 drives act like one giant drive, which may be useful if you’re using really gigantic files).
But as noted, RAID is NOT a substitute for backups. Not least because if you accidentally overwrite or delete a file that write action will be propagated over all drives immediately. Some RAID configurations might help against failure of a single drive, but god help you if you have another failure before the replacement drive is repopulated. Striped-only RAID configurations actually increase the risk to your data, and in any case adding drives will increase the risk of a drive breaking.
Backups should be made periodically and on a physically separate medium that is stored away from the source in a safe place. A simple external USB hard drive should do it. Maybe two if you have more than a TB of data.
Oh: and it’s not a backup if it’s the only place your data is stored. Backups are copies
RAID is for robustness and/or speed. It is not, in and of itself, a backup solution. Backups involve making independent copies of the data and storing them separately from the “live” data.
Sorry for your loss. Even with a full backup elsewhere I would be very upset if my Drobo suddenly stopped doing what it is supposed to do best. Those things cost hundreds of $$$$.
However, unless you have at least two complete copies of your data in two separate places, you don’t have a backup.
Your Drobo is a very reliable external hard drive, but it isn’t a backup.
What would have happened had your machine gotten a virus that simply erased the entire Drobo? Such a scenario wouldn’t be the Drobo’s fault, but you would be screwed just as bad.
See that’s the thing, I can live with a virus. I can live with a fire or a flood. These are risks I was willing to accept. If a virus destroyed my data, well that’s my fuckup for not getting virus protection software. I bought a Drobo SPECIFICALLY TO PREVENT HARDWARE FAILURE, and it died because of hardware failure. I guess I learned the hard way that its no substitute for a backup, but I still don’t see the real point of running a drobo if I have to continually back things up as well.
Well, a well-configured RAID array will let you just swap out dead disks without doing a full restore, so it increases your availability and increases the mean time to failure of the array as a whole (because more disks have to simultaneously fail for it to become inaccessible). It’s far from a panacea though, and shouldn’t be considered to be bulletproof by any means - there are lots of factors that mean you don’t achieve the theoretical maximum reliability improvement. It’s not unreasonable to back up to a RAID array, but as a primary storage system I wouldn’t treat it as too much more reliable than regular disks. This is particularly true if you’ve filled your array with consumer-grade drives.
I’ve started taking a more conservative attitude to what needs to be backed up; if you can pare down your backups to the order of tens of gigs or so, and have a whizzy internet connection, then online options like Amazon’s S3 storage service become viable. If you use a front-end like Jungledisk, you can map your online backups as local drives, and can encrypt all your data for peace of mind. I’ve got about 8GB backing up nightly, and it’s costing me less than a dollar a month in fees. For this I get very well-hosted, massively reliable storage, and best of all, I don’t have to administer it.
You’re never going to get your music/video collection on S3, obviously, but backing up that amount of data is always going to be a pain; probably investing in a Blu-Ray writer might be the best bet for that sort of archival.
Anyway, having just had my main development PC die this Monday (motherboard, though, no data loss), I can certainly sympathise. Hope you haven’t lost anything irreplaceable…
It’s to keep data live and accessible in the case of failure. A hard drive can fail and you can still access your data without doing anything, just replace the bad drive when you get a chance. I had an iOmega terraserver, RAID5. It’s completely fried now. Luckily, I backed up all the data on the RAID5 onto other disks. (Note to everyone: do not trust anything iOmega makes. I’m a complete idiot and have only myself to blame, as I’ve had two iOmega Zip drives experience the “click of death” and wipe out data on several Zip disks. I decided to give iOmega one last chance. Stupid me.)
After researching everything, the best solution for me was simple mirroring. The fewer levels of complexity, the better, I figure. If my DROBO fails, I don’t know if I would even be able to read those disks without another Drobo. Is it a RAID5 array? Is it something else? I need something that I could easily plug into any computer to retrieve my data. If I have all 4 Drobo disks (or however many I have in the Drobo), can I get at that data easily without the Drobo itself? I doubt it. I keep one copy of data on an internal drive, one copy of data on an external drive that I usually keep unplugged, and a compressed version of some of my data on a server. I haven’t had an internal drive fail yet, but I’ve had one external bite it in addition to the Terrserver. When that happens, I just run out, buy a new drive, copy the data over, and I’m good to go.
Yup, that’s the conclusion I came to as well, pulykamell. I just want to make sure no one makes the same mistake I did, because Drobo sure goes out of their way to imply that if you just buy it your data will be fine! Uh uh. It is a $500 paper weight.
As others have said RAID is for redundancy and speed. There are several levels of RAID.
As an example, the 3 most popular levels are:
RAID 0 - Striping
All data is striped across all disks in predetermined bit lengths. This provides speed. Any hard drive can only handle a certain number of I/O requests. Once this I/O limit is reached then that become your bottleneck. Striping across 2 or more drives in equal amounts per drive give you more available I/O (assuming your controller can accommodate the needed I/O requests)
RAID 1 - Mirroring
Everything written to disk A gets written to disk B. This is nice if you suffer a physical hard drive failure, then you can revert to the mirror. There may be minor config changes needed to get the second disk in the mirror to be functional, but on modern RAID controllers there may not be any config needed.
RAID 5 - Striping with parity
Similar to RAID 0 except 1 drive is used for parity. The parity bit is spread across all drives in the array in a sequential fashion. This RAID config allows for acceptable speed most times, due to the striping…but also allows for a single disk failure due to the parity.
You also have mixtures of RAID levels, such as RAID 0+1, as an example. Each RAID level has it’s advantages. Your particular needs, be it speed or redundancy, is what dictates which RAID level you go with.
Raid 1, 5, 10, and 6 will protect you in the event of the hard failure of one (or possibly more, depending) physical disks, which is actually fairly uncommon given the usage patterns of the typical pro-sumer machine and the MTBF of modern disks.
For a typical workstation user, physical disk failure is far less likely than either:
a) Filesystem corruption, or
b) Accidental file deletion
RAID won’t help you in either of those situations, since, by definition, it automatically replicates any sector-level changes to all disks in the array (and/or the parity blocks, etc).
If your main objective is data preservation, nothing beats an LTO4 tape drive or some kind of remote point-in-time recovery system. However, both of those options are hellishly expensive for home users.
So do what I do with my laptop: One external disk for daily backups. A second external disk for weekly backups that I otherwise keep in a drawer in my office. At most, I’ll only lose a week’s worth of work. Which would suck, but it’s a hell of a lot better than losing everything.
