Odd title, you say? It was inspired by a Cybernetics whitepaper that I read this morning entitled "The Risks of a Disk-Only Backup Strategy." You can read the whitepaper yourself by following this link. While I actually have a fond place in my heart for Cybernetics, I think this paper isn't worth the paper I didn't print it on. Click "Read More" to see why.
I've actually got a fond place in my heart for Cybernetics . They were actually the first company I bought tape drives from. They had some cool technology, like a device that could create two simultaneous copies of a tape to two mirrored tape drives. Very nice.
But the more I read this paper, the more I have to disagree with it. The basic gist of the paper is that a disk-only backup strategy isn't good because disks break if you drop them (more often than tapes do), and they "prove" this point by performing a series of drop tests on tape and disk media. While I don't argue with the tests themselves (much), I argue that the way they did the test made the results irrelevant for our purposes. I also argue that there are a number of points they didn't make that seem rather obvious to me.
- Almost no one carries their disks offsite
- They make the point that disk fail more often than tapes if you drop them from the height of a desk. So what? Almost no one who is talking about a disk only strategy is talking about physically carrying a disk anywhere! So what does it's ability to survive a fall have anything to do with anything?
- If they did carry them offsite, they would probably use a mobile disk drive
- They are vendors (like Prostor) that have designed tape-like media that is based on disk, but that can be handed to an Iron Mountain person. But almost all of them intended for SMBs or enterprise are built using laptop disk drives, not 3.5" disk drives like what they used int the article. Laptop disk drives are designed for much more stringent requirements, especially with regards to surviving being dropped.
- Any disk drive being carried offsite would also be in a container designed to do so
- They did the tests with a bare disk drive. Give me a break. Even someone using a USB disk drive would have some kind of shock protection around a disk drive. No one is taking a bare disk drive and handing it to their Iron Mountain driver. If they are, they're an idiot, because it just wasn't meant for that. And if you drop it, it breaks. Surprise!
- 3 of the DLT tapes failed out of the box
- They mention but completely gloss over this event. 3 of the 10 DLT tapes they bought failed out of the box! That's kind of the point that pro-disk-drive people have been trying to make. Disk drives do fail, especially if you drop them from 4 feet with nothing around them. But tapes fail all the time with no good reason as to why.
- The test was too small, and anecdotal evidence is different
- How were these tapes and disk drives dropped? Did they try different kinds of drops, like corner, side, back, etc? It's not possible that they did considering the number of tests they did. Speaking of which, even if you ignore the different kinds of drops and just use one drop, the number of drops they performed wasn't enough to be considered statistically significant. Then let's look at the anecdotal evidence. I've personally accidentally dropped a number of tapes in my time, and many of them failed. Springs popped out inside the device, doors popped off the hinge. You name it. So don't tell ME that tapes can withstand shocks better than disk drives.
- Disks fail as they get older. Duh. So do tapes.
- They cite the Google study, which was very interesting, to point out that disks fail more as they get older. Wow, there's a news flash. Have you looked at an old tape lately? I know plenty of people that stop using a tape after only a few dozen writes, due to their experience in older tape reliability.
Don't get me wrong. There are risks associated with a disk-only backup strategy that are not present in a tape-only system or a hybrid disk-tape system. This paper just doesn't disk any of them. Maybe I'll write another blog post about those risks.
----- Signature and Disclaimer -----
Written by W. Curtis Preston (@wcpreston). For those of you unfamiliar with my work, I've specialized in backup & recovery since 1993. I've written the O'Reilly books on backup and have worked with a number of native and commercial tools. I am now Chief Technical Architect at Druva, the leading provider of cloud-based data protection and data management tools for endpoints, infrastructure, and cloud applications. These posts reflect my own opinion and are not necessarily the opinion of my employer.