Dave A. Anselmi

Digital Filmmaking & Post

There is a tremendous amount of bad information and misinformed opinions on the net regarding the longevity of hard-drives. Is an external drive a robust archival solution? Many casual users seem to like the low-cost solution in all things, regardless of risk, and will post glowing reports, even if they’ve had a failure or two.

Correspondingly, as professionals, we’re always balancing risk vs cost. Thus, if the chance of catastrophic failure doesn’t bother you, then cheap makes sense. If your business model doesn’t have a lot of clients returning years later for re-use of previous footage, then ‘robust archival’ perhaps isn’t as crucial to your business. I like my biker friend’s perspective, when asked if carrying a patch-kit & pump is worth it on long rides. He answers simply, “How far are you willing to walk?”

Here’s some info I’ve gleaned, working in a high-tech company at my day job:

First: MTBFR or “mean time between failures” means, if you’re lucky, you’ve above the mean. If not, you’re not. Consider the old saw: if Bill Gates and you were standing in the elevator, the average salary of the elevator occupants would be millions of dollars, but does that change your salary? “Average” and “mean” are different things, and in this case, it makes my point even stronger. ‘Mean’ doesn’t indicate that ‘results trend towards’, it just indicates ‘equal amounts of hard-drives failed before and after that time’.

It’s next to impossible to rate hard drives by manufacturer. In the past, many drive-makers were OEM’s obtaining product from a single manufacturer such as IBM. You could compare which design was more robust. For example, at one point IBM 3-platter units worked well, but 4-platter units had problems, etc. Today the job of comparing drives is more difficult. One model can be made in different plants. And the same model built in the same plant can have different results over time. This is because the profit margins on drive hardware are so low. All manufacturers are cutting costs. It’s telling that IBM no longer sells consumer hard drives.

Here’s an easy test. Go find your selected drive on NewEgg. Even if it’s 4 1/2 stars with 100’s of glowing reports, read the “1-star” reviews. There may not be all that many, but they are important when they report catastrophic failures. Can your business withstand a catastrophic failure, if your drive’s number comes up?

For casual video editing storing your footage on hard drives is fine because casual use doesn’t work the drives for hours on end, day after day. Correspondingly, there really are a plethora of horror stories on the Avid-List of people building RAID-0 striped arrays (For an explanation of RAID technology see which worked so well for years, until they crashed & lost everything. At which point, most swore off the ‘cheap route’ and bought some kind of redundancy e.g. RAID-1, RAID-5, RAID-3, etc. Or if you need speed, use RAID-1+0 or RAID-0+1. These latter RAID systems are called “Enterprise Class” drive arrays which are the systems running Amazon, eBay, and other large e-commerce sites.

Finally, let’s look at archiving to external drives e.g. drives in removable caddy’s or external drives you unplug and stick on a shelf. Most professionals agree that drive bearings need to be spun-up at least once a year. If your drives are being rotated off the shelf yearly, then you’re going to tend to have much less failures than someone who just stacks them up there for years. But for true archival, tape is still the best. Flash-based media is a complete no no for archival backup, regardless of what marketing departments say. Their catastrophic failure rate is so much higher than hard-drives that manufacturers refuse to post it.

The same is true with optical media such as CDs and DVDs. No one tries to sell optical media as an “enterprise ready” archival solution. Enterprise companies would jump on the chance if they could. But they can’t. Yes Dorothy, in 2011, we’re still relying on rust on scotch tape simply because you can play futbol with a LTO-3 tape (Linear Tape-Open), toss it in a pond, let mold grow on it, then bake it, air-blast it, and still recover everything.

Right now the sweet spot is LTO-3. If you really need bullet-proof archival, it’s probably best to get an external LTO-3 drive, and hook it to a dedicated computer, so it can cache the data. You don’t want to drag and drop to any kind of tape drive. This will back-up extremely fast, nearly hard-drive speed if extremely noisily.

For the kind of daily video-editing we do, I highly recommend something like a Nitro-AV dedicated RAID unit e.g. RAID-1 or even RAID-5. They’re the OEM for I’ve used them for years, and when drives have failed, they recovered transparently. Even cooler– they support RAID-3, which is optimal for video editing (the stripes are larger, and the parity stripe doesn’t rotate across disks), and is what Autodesk (Discreet) uses for Inferno, Flint, Flame, Fire, Smoke, and those multi-million dollar systems.