Will your Drobo keep your data safe? Here’s how Drobo keeps you from knowing…
I’m now on my second data loss situation in three months: my Drobo time machine drive has given up the ghost with a corrupted directory and no longer mounts in the Finder. Very frustrating, especially since I have too much media to be able to use any traditional online backup utility like Mozy or Carbonite. Fortunately, I have backups.. But why should I need them? None of my physical drives have been damaged, so shouldn’t my data be there? Drobo has shifted risk away from hardware failures. But the compromise is you now have to trust its complex drive operating system to manage your data. Its an unquantifiable risk that works to Drobo’s advantage.
So why use Drobo? Drobo Inc has always impressed me with their product value proposition, even though I have experienced multiple data loss and drive firmware incompatibility issues with their products. This idea that I can just have a self managed, self-healing RAID system available to store my 4 TB of photos and media is sexy. I have three Drobos, spread out over three Macintoshes. I don’t trust any single one of them, because I’ve frankly had too many glitches with their products over time to completely trust my data to their drives.
Of course, an important part of the company’s image is their ability to promote extreme reliability. How do they do this, when there are so many anecdotal reports of Drobo’s failing to keep data safe? In older days, major user problems with a data storage product would be extremely well reported because there were only a limited number of forums where users could discuss this kind of problem. There were also less products out there. Today, I think the long tail has created a situation where its much easier for a savvy company to preserve its image against most moderate user complaints. The fact is, that pretty much only bloggers who bother to write about their negative experiences are going to get any search hits on Google. And most bloggers have better things to do with their time unless they are very frustrated like Fraiser Spiers, or a few others. And product review sites like CNET let a few people air sour grapes but don’t provide truly comparative information about reliability.
So is Drobo a safe place for data? I don’t know. The only data loss situation I have ever personally faced that wasn’t the result of an actual hard-drive failure was from my Drobos. Probably the only people who do know the real situation is Drobo since they can mine their support database in a way that an outsider cannot.
I have used more than 10 large single and RAID0 and RAID1 drive systems to store many TB of data. I’ve experienced single drive failures that have lost data, but that is due to HARDWARE failing, not SOFTWARE. In my 20+ years as a computer expert, I can tell you the reality is that typically only a massive hardware failure is going to screw up your directory beyond the normal repair of your Windows or Macintosh file system. Both operating systems use filesystem journaling, so are very resilient except in the case of your actual physical drive or controllers failing.
So if most data loss occurs due to hardware, not software, failures, Drobo sounds great on paper. The reasoning goes: If you could eliminate the risk of the drive itself failing, like Drobo claims to do, you should have a very safe system, right? No. The inherent strategy that Drobo uses in its BeyondRAID strategy, is effectively a “second layer of indirection” around your data, even more so than a regular RAID scenario because it is “volume content aware” and must constantly and dynamically reallocate its redundancy formula across disks. I believe most RAID systems use a far more deterministic, and therefore less error prone, strategy to figuring out how to lay your data across multiple drives. Therefore, it wouldn’t surprise me to learn about fairly widespread problems with the Drobo product.
Drobo has shifted your risk away from single hard drive failures, but increase the risk of failure due to defects in its own software-based firmware. But how risky is it? No one knows, and thats good for Drobo.
Strategies that Drobo uses to contain its user-base from this kind of information:
- Admit right up-front that your products are not necessarily safe in your FAQs
- Maintain an “Apple”-like image at all time with your product value proposition
- Create an online forum you that is strictly policed where there is a trickle of content from happy users, but nothing from unhappy users; Only allow users with a valid serial number to access your online forums
- Populate your own company web pages with heavily linked pages using the same keywords that frustrated users would use to search for problem reports (Example, Example 2, Example 3)
- Carefully script your tech support people to shift blame away from your product in a very friendly and non-antagonistic way (”data loss can happen with any drive, blah blah blah.”)
(Some examples of user problems: example, example, example, example)