Am I stupid to buy OEM Hard Drives?

December 9, 2008 1:47 PM

I need advice on the reliability and economics of buying OEM Hard Drives in bulk. Is it false economy?

I work for a video & film production company that generates a ton of data every month, now that tapeless video acquisition is the norm, and not the exception. We used to go out to BestBuy every so often and pick up a cheap external Firewire or USB drive that we'd archive our project data off to and the end of a job.

But it quickly became apparent that the external FW/USB drive route was clunky and inefficient. The varying form-factors of the drive enclosures and incredibly annoying powersupply juggling (we'd use bus-powered drives if they were big enough and fast enough to dump all our data, which they generally arent) caused me to investigate other options.

I came up with a plan to buy a NexStar FW/USB Hard drive dock for all of our editor's workstations, and buy bulk 3.5" 1TB SATA bare drives that they can just plug into the docks and dump off the data. Then we'd put the drives into our media vault using something like these Wiebetech Protective Harddrive cases (or a cheaper, reasonable facsimile).

(We backup all this data to LTO-3 tapes, as well, for redundancy. I trust neither tape nor spinning platters of glass & metal on their own, but together, I think it's as solid a backup plan as our budget allows for)

So now the question is, what hard drives do I buy? Since we generate at least 2-3 terabytes of backup data a month, I think we'd have to buy the drives in bulk in order to get any level of economy from this plan (preferably in lots of 10 at any single time, in case drive prices fluctuate). I noticed that on sites like NewEgg.com, the prices for "OEM" drives are significantly cheaper than their retail boxed counterparts. That's fine with me, as I dont need any of the screws, brackets, cables or software that the retail units ship with.

What's disconcerting to me, however, are the seemingly large number of negative reviews about these OEM drives, in particular, this Seagate Barracuda 1TB drive (I always buy Seagate by default, mostly out of superstition, but also because I've had many more memorable drive failures with Western Digital and Maxtor drives. But I'm more than willing to be convinced otherwise).

Is this a classic case of "only people with problems post product reviews on the internet", or are these OEM drives really inherently sub-par to the retail versions?

Any advice, info or pointers to better bargains are most appreciated!

posted by melorama to computers & internet (11 comments total)

I've personally never encountered a situation where a drive marked OEM wasn't physically identical to a drive you'd pull out of a retail box. I also doubt Seagate wants their high-volume customers (system builders, etc) to be put off by an inferior product.
posted by Rendus at 2:16 PM on December 9, 2008

OEM drives, bought from a respectable marketplace, should not have any problems that a full retail drive would not also have.
posted by shmegegge at 2:25 PM on December 9, 2008

I buy OEM drives from NewEgg all the time and I have not had any more or less problems with them than boxed retail drives. As you mentioned, it's just less junk to throw away when you get it.

(Also, thank you so much for linking to those Wiebetech hard drive cases. I've been looking for something like those for so long.)
posted by joshrholloway at 2:30 PM on December 9, 2008

As far as I'm aware, there isn't any difference in the drives themselves - just the packaging and accessory content (cables, manuals 'helpful" software, etc.).

I almost always buy OEM drives and haven't noticed anything that makes me think they're any more or less reliable than retail drives.

Of course, If I was in your situation, I'd probably buy a lot of OEM drives, a SATA II raid controller from a reliable brand (think adaptec*), a solid motherboard and a large case. I would then couple a large raid 5 or raid 10 array with a Linux or BSD based NAS-focused OS and plug it into the network. Back it up with tape and call it good. Need more space? Build another box or add another array to the same box.

*When doing this stuff for important work, I'd suggest buying an identical raid card to keep as a spare. Good to have on hand - not all data issues are related to the drives alone.
posted by terpia at 2:31 PM on December 9, 2008

OEM drives are just bare, not inferior in any way, however warranties offered may be different. (YMMV)

You probably just came across a poorly reviewed drive, and it just happened to be OEM.
posted by wongcorgi at 2:34 PM on December 9, 2008

Any reviews of an OEM drive are reviews of a particular model which, as Rendus says, is in all cases I've ever seen the same drive whether it was sold in minimal packaging or boxed up all pretty with a nice poster. Whether the reviewers believe that or not is a different matter. It's purely that the stores demand something packaged as a high-end consumer product, and the rest of the market (repair and upgrade businesses, small PC-building enterprises, technically-inclined home or office users) just wants a good product at a competitive price.

Whether that particular Seagate drive is bad I can't say, but I've certainly had a couple of Seagate drive fail very early. Don't be afraid to consider Samsung, Fujitsu or even IBM. I've had very reliable drives from all three.
posted by le morte de bea arthur at 2:41 PM on December 9, 2008

I did something a lot like this for an experiment that took data at (up to) about 100 GB/day. Our primary data storage and access was a set of off-the-shelf RAIDs. We also had some key-removable IDE drive bays. As the data came in it went both to the RAID and to the IDE disk; a couple times a week, someone had to power down the machine to swap the single disk out. This setup let us recover pretty painlessly when a power supply on one RAID controller caught fire and we lost sixteen "redundant" disks at once.

This was a couple of years ago when drives went for $0.50/GB. Today Newegg's "internal drive" page lists several drives under $0.12/GB. The great big Seagate you're anxious about is $0.10/GB. I wouldn't get it yet; maybe you'll learn something else next month.

We bought our drives a dozen or so at a time, every few months. They came to us in nice egg crates. Also: if you're going to be filling a drive every week, go the the sewing craft store and get a nice label maker :)
posted by fantabulous timewaster at 2:45 PM on December 9, 2008

My (very) limited knowledge seems to indicate that the drives on the bleeding edge of massively large are sometimes unreliable in their first incarnations.

BTW if you are buying a lot of drives, it might be a good idea to monitor slickdeals.net, they list great buys in hard drives pretty often.
posted by sully75 at 3:42 PM on December 9, 2008

The only thing I would be concerned about is whether the vendor for the OEM drives is handling them sufficiently carefully. I doubt it's much of an issue with places like Newegg, but I have in mind the way they're displayed if you go in the Microcenter near me - all piled into shelves in no more protection than an anti-stat bag. You have no way of knowing if some moron was juggling them - badly - an hour before.

That doesn't stop me from buying them, but I don't have your concerns.
posted by phearlez at 4:01 PM on December 9, 2008

phearlez mentions worry about whether drives are handled carefully. Even when I was filling a couple of drives a week, I never had a drive failure that didn't happen during normal operation. In particular we never had any electrostatic-shock failures, and we were operating in a pretty dusty environment in a low-humidity climate.

Remember that drives are manufactured products, and some fraction of them will always fail, no matter what. The whole point of backups is that you don't care whether your storage fails, just how often. You're thinking of buying, filling, and storing dozens or hundreds of drives in a year, which gives you a different sensitivity to failure than most people:

• In 2007, the Seagate company shipped something like a hundred million units a year, at an average of $70 apiece. If they had a 0.1% failure rate that'd be a hundred thousand RMAs a year, or 300 every day; this kind of workload would require a large, but not absurdly large, department in the company. If the company-wide failure rate were were 0.01% they'd get 30 calls a day, which you could probably handle with a couple of techs and a secretary. For the manufacturer, the difference between 0.01% and 0.1% is big money --- maybe few million dollars a year, a couple percent of the company's net income. OEMs putting the drives in new computers work at the same scale, but have to deal with more types of components, and have correspondingly bigger tech support costs.
• Most consumers buy a drive every year or two and are completely fucked if/when it fails; most of the failures come after two or three or five years, at the end of the drive's normal life, after the drive has accumulated a lot of the consumer's valuable data. These consumers usually know somebody that had a drive fail (early or not), and that person had a lot of trouble, but they don't have enough information to learn how to tell an intrinsically bad drive from an okay one. Looking for patterns, without enough data for real patterns to develop, tends to lead to superstitions. It sounds like you are here.
• A storefront or big-box retailer might move a few drives a day, maybe a thousand or ten thousand in a year. But for the most part the retail guys don't actually install the drives in anything, so they're not sensitive to whether the drives fail before the customer loses his receipt. For a big-box store with a "geek squad" that installs three drives a day in customers' computers, a 0.1% failure rate is one botched install a year; a 1% failure rate is a botched install a month. In a store the right size to put a handful of unboxed drives in the window, juggling the drives is basically the only way to produce a noticeable number of problems.

You're thinking about filling a dozen drives a month. If I were in your shoes, I'd start off planning for 2-5% of them failing early --- that way you're covered when your doofus coworker tries to carry three and drops them. If you get through a hundred drives with no failures, you can say with some statistical confidence that a 2% failure rate was an overestimate and budget less for it. Without specific evidence for static problems, I would hang onto the antistatic bags from shipping rather than buy each drive a $5 antistatic box. Plus, with transparent bags, each drive only needs one label.
posted by fantabulous timewaster at 12:52 AM on December 10, 2008

Nope. Not stupid at all.

Drives die. Google has some cool trends on their datacenters, and they realize that it's not the manufacturer...it's that they die quickly...or make it to the 2+year point...with consistent failures...and then the number starts to get higher.

On video: a Raid 5+0...for 'long term' storage..use two drives.
posted by filmgeek at 7:37 PM on December 26, 2008