Here's why you should not buy a Drobo
jan.rychter.com
jan.rychter.com
The "reliable" data storage device I bought corrupted and lost my data once, and right now is showing the symptoms of corrupting the filesystem again.
I doesn't get any simpler and more objective than that.
And to those people whose Drobos have been working fine so far: I wish you the best of luck and I hope you won't hit the same problems I did. But I would be worried, beause if I'm hitting them, it means they are there, and eventually you might as well.
The point here is, Drobo/RAID/NAS are disk redundancy solutions - they cater for the fallibility of spinning magnetic platters, and never promise 100% reliability. Just like any solution, if you want to avoid data loss, you need to have an orthogonal backup system for when the other fails at a system level.
It sucks that he's had such problems with his Drobos, and it looks like the support he's got is subpar. This is not, however, the universal experience. Based on my experience, I'd be happy to recommend the Drobo to most people I know who want a simple and easy to setup solution for hard drive failures, with the big caveat that they still need at least one other backup solution.
Tip: don't buy the old drobo, get the new one. Don't get the drobo share, use an airport extreme or similar disk-sharing network device.
A one device solution? Well, that'd be a router, switch, hard disk... (can you say, computer?)
sometimes we want stuff that's single purpose, done well, easy on the power/heat/etc. I now have an appletv on a turn-off power strip, enabling me to 'turn on' the tv system and have something running easily, rather than having to get my laptop out, an external disk, try and make it all work. Ugh. madness.
Still, after seeing the corruption and after trying their support, I don't think I'll want to buy any more Drobo products ever again.
It is a bit unsettling that Drobo advertises these are completely fool-proof in their marketing materials but support, rightfully I may add, recommends routine backups. In the end, it's a JBOD RAID in a pretty enclosure with approachable software. For me, that works.
All I wanted out of my DroboPro was a way to combine eight drives and have an expandable dual-disk redundancy array. It's working like a champ for that.
My drobo with all WD green drives is as silent as my Mac minis and MBP are, while a friend of mine has one also with similar drives that's so loud it's very difficult to sleep in the same room with it (and yes, it rivals his Mac Pro in terms of noise).
I've not had a single problem with my drobo since I got it almost two years ago, while others keep having them fail repeatedly and replaced repeatedly until they give up and find something else.
I've also come across people that think their drobo is great and infallible, and then go and realize they have no backup for the file they just deleted off their drobo. And come across people like me that back up their drobos because a drobo is purely redundant, which doesn't cover cases like accidentally deleting files off it and losing the drobo to fire/flood/theft and more, only situations like a dead hard drive or two.
I love my drobo to pieces and recommend it to anyone that wants a solution they don't have to manage...with the caveat that it's their data and therefore their responsibility to take care of it, and a drobo can only be one part of the solution. data robotics customer support definitely has a bad reputation that is probably warranted to an extent, but not many NAS manufacturers, nor you if you roll your own solution, can do much when things go very, very wrong with your device for whatever reason.
(Note that I use my drobo only as direct storage. The drobo NAS solutions always seemed a bit hacky to me...)
Buffer I/O error on device sda1, logical block 422806282 scsi2 (0:0): rejecting I/O to dead device
SCSI error : <2 0 0 0> return code = 0x70000 end_request: I/O error, dev sda, sector 4533105544
HFS+-fs: WARN: Turning journaling back on even though Journalling was not on at mount. HFS+-fs: WARN: We may be setting the Journal bit on an unjournalled filesystem but it is OK.
So please don't soften the issue — it fails. It failed for me multiple times already and I expect it to fail again.
I guess it's OK to use a Drobo if you get another one as backup for the first one — but then, what's the point?
RAID is _not_ a backup. If you don't understand this, you'll never be satisfied with any RAIDish products.
RAID should increase reliability, not decrease it. Any RAID solution that results in a net increase in data loss is not a good solution.
If I back up bad data, I have older but good data to recover from. If my drobo catastrophically fails, at the worst I've lost data that is probably not going to be difficult to replicate/replace because of how new it is. I don't have to start ripping my hair out at the idea of having lost an incredible amount of data ranging from irreplaceable photos of my life to work and more.
In the ideal case, raid should increase reliability, and they do, only for one definition of reliability. But freak accidents and poor design happen, and it's silly to rely only on a raid for reliability.
If my drobo fails, so be it, drobo sucks etc., but my data is my responsibility. I have backups in place to deal with the data loss. I'm not asking my drobo to be perfect, I'm just asking it to keep on working in case a drive starts to fail, but there's clearly many possibilities where drobo itself fails. If there's fs corruption going on, it's my responsibility even if it's the fault of the hardware. If there's something I deleted off my drobo by accident, it's my responsibility.
Now, if you were going on about the bad design and hacky nature of droboshare and the software drobo runs on, and data robotics' bad customer support, then sure...you have a point. We could all learn from your experience not to use droboshare. That's about it.
When I was shopping around for drobo alternatives (back when I couldn't stomach the price just yet to buy one and I was getting a bit sick of my hacked-together raid in one of my computers), product reviews and blog entries read exactly like what I see here with the drobo being hit or miss. There were people that were all "DO NOT BUY!!" because they lost all their data next to the people just gushing over how much they liked theirs cause it was working like a charm. This is a tough market, because people don't want to lose their data, and when (not if...) they do they get quite pissed.
It's really up to you to determine your needs and to find a device or twenty that meets them. I rather like the drobo because it's zero maintenance. I don't trust my data to it, I only trust it to keep working when a drive has failed. But that's about all I can expect from any such device. The whole market is pretty much hit or miss for people - it's a hit until it misses, then all you can think about is how it missed.
When it works, it's fantastic. When things start to go wrong, there's nothing you can do about it except experiment with switching drives or email support and hope they can diagnose it.
I loaded my first Drobo with Seagate 1.5tb drives. It constantly dropped drives out of the array and was very slow (<10MB/sec over FW800). After several back and forths with tech support I gave up in frustration and didn't use it for a while. Then 2TB drives became available, I swapped them out, and it stopped dropping out.
I then upgraded to a DroboPro as I had outgrown the 4 drive version, which was painless. It was still slow, but not as much. The log is now ENCRYPTED(!!!) so there is absolutely nothing you can see for yourself other than whether a drive is "bad" or "good". If you suspect that one of the drives might be failing, you have to email your encrypted log to Drobo and wait for them to respond. Which is pretty frustrating when you're in another timezone and each exchange takes 24 hours.
Somewhere among switching around drives and upgrading firmware it became faster, and now to me seems acceptably fast. I think I must have just had a bad drive or three, but of course without access to any logs it's a tedious and expensive process of trial and error to figure out.
All was well, and then just shy of a year it stopped booting. I bought it in one country and then moved, so had to ship it back to the original distributor, which is annoying and expensive but understandable, but of course you can only access the drives with another Drobo, and means I'll be without access to my stuff for about a month (except for backups). I wasn't too impressed about what is marketed as a professional product designed to run 24/7 failing after less than a year of light use.
Pros:
* Very easy to use. Auto-expansion is fantastic and works as advertised. You literally only have to shove a new drive in the slot and wait a few minutes. * Don't have to rebuild when you add capacity * Don't lose all your data if you disconnect 2 drives at once, put them back in a different order, or any of the myriad interesting ways you can accidentally kill your entire array with RAID.
Cons:
* Somewhat pricey. But you get what you pay for - at least I did until it died. * Encrypting the log files is outrageous. This is probably the worst thing about it. * No transparency about what it's doing. It's quite literally a black box. * iSCSI on Mac doesn't reliably reconnect if you unmount it. I usually have to restart the device and/or my PC. * No way of accessing data (other than another Drobo) if the device fails. * Not particularly fast. Much slower than a bare drive. 40MB/sec if I'm lucky. No idea how they got VMware certification; trying to boot from it would be agonizing.
Other thoughts:
* Others have said it sounds like a jet engine. It's not quiet, but it's by no means loud. It's just two 80mm fans spinning at maybe 1200rpm or less. Drives always seem cool to the touch. You probably couldn't sleep with one next to your head but in the next room, no problem. * It takes days to rebuild 8TB when you switch out a bare drive. So does normal RAID though.
I'm a big fan of either building a linux-based server, or using a QNAP. QNAP is a bit pricey per disk and per GB, but it just works.
For large disk requirements, I use a linux server with Areca 12-port RAID as secondary storage, and external eSATA+expander trays for tertiary storage (probably could put 50-100 drives on this with no problems). Overkill for most home use, though.
It's 2010, and I'd use iSCSI, not SMB, unless I wanted direct-attach FW800.
The other issue is that write (and sometimes read) performance is hurt during rebuild. For timing-sensitive systems, people often turn off auto-rebuild or scale back rebuild percentage of writes, so their application's performance isn't so affected. Of course, if the second failure happens during the rebuild...
(Incidentally, Netgear UK are doing a "free 2 x 500GB drives" promotion until the end of June: http://www.netgear.co.uk/freedrives_nvx.php)
Until an equivalent OEM product comes along that supports something like ZFS, the ReadyNAS is probably one of the best options available.
AFIAK the only way to damage a zpool is if some snafu (one more serious than a mere write failure) wiped out a bunch of metadata together with the backup metadata all at the same time, but such an event would do very serious damage to any filesystem, which is why all important data should have an off-site backup.
You can then run "zpool status" and it will display the last time each pool was scrubbed, and show you the number of errors if any.
First, I'm sorry for the troubles the original poster (Jan Rychter) has had. I don't have any first hand knowledge of his use, and can't find a support case on it -- has it been escalated by our European team back to the factory? Please send me your case # (via direct message at drobospace where you are a member) and I will escalate if necessary.
The Drobo/DroboShare combination has a binary user experience. Those with no prior network storage experience love it, those with prior experience want more functionality and performance. Providing more performance is the goal of the Drobo FS, a single box solution for sharing via AFP or SMB. Third party DroboApps provide web, ftp, rsync, NFS, and other protocols. This box is 4x faster than the Drobo/Droboshare owing to eliminating USB as an interconnect, and faster generation of ARM processors.
Mark Fuccio
1. Regarding the fan noise - I used to keep my Drobo in a warm house with no air conditioning. Temperatures in my office would regularly reach 78-80 degrees, and anything in the high 70s would cause my Drobo fan to go into "vacuum cleaner mode," which is really annoying. Solution: Don't keep your Drobo in a hot room. This is a sensitive piece of electronic equipment with lots of spinning platters stacked on top of each other. Drobo is more sensitive to heat than other pieces of technology. Since then I keep my office air conditioned and at 72 degrees, the Drobo is silent other than the hard drive seeking itself.
2. Filesystem support - This is a problem with any "cross-platform" storage and the fault is not Drobo's. I use my Drobo on both an OS X 10.6.4 system and a Windows 7 system. The only filesystem I can use that supports both seamlessly is FAT32. Unfortunately, FAT32 doesn't support some character sets that Mac does, which I found out when I was trying to back up my iTunes library there. Another limitation of FAT32 is that your filesystem can't be larger than 2TB, and you can't store files larger than 4GB. My other option was to go with HFS+ or NTFS but either of those solutions are Mac or Windows only.
The two main issues he has: Heat and filesystem compatibility, can be solved by simply using the Drobo differently. Drobo is not without it's limitations. The speed is slow, and it's sensitive to heat. But if used properly, it works fine as a large data storage device.
My workload is pretty light though as it hosts media and I only stream it from two devices max at a time.
But it's hard to beat the price and comprehensive protocol support. FTP, Samba, NFS, AFP, Time Machine backups, iTunes sharing, etc.
Have you anyone tries to use freenas as a time machine destination?
As for power consumption, I unfortunately haven't kept tabs on the electricity nor have a Kill-a-Watt.
But I have told FreeNAS to pretty aggressively spin down unused drives.
I've been planning an upgrade to 4x2TB for a 6TB RAID, but I haven't had the time or the $ for the upgrade.
I used some of the storage space as a time machine destination, but Apple has issues with Time Machine over SMB, and it got to the point where it was too slow, so I bought an USB disk and just back-up to that.
1) Encrypted logs. I think it is crucial to have readable log files to be able to understand what is happening with the drive array. If there is proprietary information that the company wishes to be protect, I welcome them to do so. There should be two files: one which is plaintext with the major status and event messages, and one which is encrypted with their proprietary information. Rolling the plaintext log into the encrypted log seems like a user-hostile posture.
2) Maintenance costs. If the log files are encrypted, then when the array fails (not "if", "when") then I will need to buy drobocare ($129) to get them to read the logfile to tell me what happened. I expect to keep this array running for the next 4-6 years, and I am not willing to buy ongoing yearly maintenance ($250/three years) just find out how the array is doing.
Like I said, I was hours away from ordering a drive, I really like the simplicity of the design, but thanks to some of the comments on this thread I suddenly realized that I would signing up for years of maintenance fees. That's not what I wanted.
I use a cron script which runs rsync each hour and synchronizes my / and /home/ with my /mnt/mirror/.
Ever since I've set this up, I've had a reliable backup which I don't have to think about. I've even migrated my system to it once, because I needed to swap the primary hard drive out.
What's the big deal about backing up? Why buy a dedicated unit anyway? Am I missing something fundamental?
If your primary drive beings to silently corrupt data due to bad sectors (I've had this happen once. Your system starts to hang, applications start to unexpectedly quit, you start hearing static/clicks in your music, photos seem jaggy even if they were taken with a 10MP DSLR.) you will back up bad/corrupt data over your backed up good data.
Best advice is to do some research, like I noticed their closed support forum too, and thought it was suspicious - it's a good sign that they let their customers talk to each other, but if it's hidden then something's not right.
Once you get to the Drobo/WHS point of your NAS life, HDD bays and xfer speeds quickly become the biggest issue. (4x2TB - redundancy) can quickly become not enough..and chances are that you'll always be ahead of where HDD storage is.
My biggest gripes with the drobo were loudness (no really, it sounds like a jet engine), inability to recover data directly off the HD if the drobo dies, horrible transfer speeds, and horrible support.
It's one pro was simplicity.
The downside is that unraid will only protected you against a single disk failure, whereas WHS might protect you against multiple disk failures - depending on which disks go.
We're increasingly using our own custom debian linux box now with SMB shares, backups with duplicity, Apache / Lighty and so on. It just allows for more flexibility.
Secondly, what does the author (or you guys) recommend as an alternative to the Drobo? Just creating your own file server/NAS? Do you need a hardware RAID card, or are mobo-integrated ones or cheaper software cards sufficient?
It turns out you should make this faux file size as large as you can (16TB) as the faux file size restricts the actual file size, so I made my faux file size 2TB and now actually have 2.5TB of storage, to make my drobo recognize it I have to reformat all the disks and lose the data. Daft.
Some companies might do this, and you might be used to getting a semi-canned response on Twitter every time you mention a company's name, but that doesn't translate into any kind of obligation on their part.