SSDs, data loss, electricity, and hype

I’m not particularly worried about this, but under the very worst case scenario, certain solid-state disks can theoretically lose data in a week or two if they’re left without power. But that doesn’t instill panic and get clicks when you say it like that.

But you knew I was going to write about it. Let me tell you why I’m not worried.

First of all, the study calls out enterprise SSDs, not desktop/laptop SSDs. They’re built to different specifications to get different performance characteristics. Theoretically, desktop/laptop SSDs could lose their data if they’re left powered off for two years under normal conditions, but I haven’t actually seen one do that yet.

According to the study, which is a collection of worst-case scenarios, if you take an enterprise SSD, designed to sit idle for a maximum of three months, then store it in a really hot room, it may start to lose data in as little as a week. Every 10 degrees or so you raise the temperature doubles the speed of the data loss.

But you have to take that at face value: a cascade of worst-case scenarios presented by someone who just happens to work for Seagate. Seagate is trying to get into the SSD business but in the meantime has a hard disk business to protect, so a Seagate employee wouldn’t be my first choice as a go-to person for objective advice about SSDs.

The other thing the study doesn’t mention is that the conditions they describe apply to any form of stored media. I remember having a discussion more than a decade ago with my then-director about the conditions a client was proposing to transport and store backup tapes, and he was looking for validation and a polite way to tell them they were asking for trouble. Tapes are sensitive to temperature and humidity, and nobody seems to be boycotting tape over that.

And we won’t even mention optical media.

Hard drives are sensitive to temperature too. Try leaving a drive you don’t care about in your garage all winter sometime and see if it’s readable afterward.

CPU life expectancy halves when you increase the temperature 9-10 degrees too. That’s why we put heat sinks and cooling fans on them. But I don’t remember anyone throwing fits about that–and I remember the time when CPUs didn’t even have heat sinks on them. Eventually, you started seeing the text “heat sink required” on the chip, and no one complained. Then you started seeing “heat sink and fan required,” and I don’t recall anyone complaining then either.

The other important thing is that these conditions are based on minimum specifications, not necessarily real-world conditions. Based on the specifications, you’re not supposed to be able to write a petabyte of data to an SSD without it wearing out, but some people have written two petabytes and gotten away with it.

Koreblog’s advice to create forensic archives of anything that’s being used in litigation is something you want to do anyway, whether you’re dealing with spinning rust or solid-state silicon. We’re still trying to solve the problem of entropy.

So, in case you’re wondering, I don’t regret my last SSD purchase, and I’ll be buying another one when I need increased capacity. My recommendation is to keep using SSDs in your everyday machines. If you have machines that you use extremely rarely, then consider platter drives in those. But I can say my travel laptop, when I was using it strictly for travel, would sit for weeks without issues. I kept it in a bedroom closet when it wasn’t in use. My netbook still has a Samsung SSD in it, and it spends months on the shelf, coming out of retirement when I need a compact Linux box–usually during massive website maintenance–and I haven’t had issues. The old, tiny 8 GB Jmicron-based SSDs that netbook and my mom’s netbook came with originally sat in my basement for a couple of years, but they worked fine when I pulled them out of that box and pressed them into duty.

Call it anecdotal if you must, but I own eight SSDs and haven’t had any issues of this sort. The company I work for has thousands of them and hasn’t had any of these issues yet, that I’ve heard of–and when there are reliability issues on desktops and laptops, I’m one of the guys who hears about it first. Maybe my experience is anecdotal but my employer’s has definitely crossed that line. If you buy an SSD from someone who makes the chips–Crucial/Micron, Intel, Samsung, Sandisk, or Toshiba–to ensure you’re getting the very best memory chips, then use it and enjoy it, you’re very unlikely to have issues.

And if I have important data, I back it up. My backup media is inherently less reliable than my primary media–my SSDs–but that’s been a problem all along. The last time I was responsible for tape backup, we had reliability issues with tape too. You need backups anyway, because sometimes it’s not the drive that failed. To me, that’s the biggest danger.

I apologize for the defensive tone, but I hate seeing people trying to stifle innovation. I remember a time, about 30 years ago, when computing was really exciting. It was the dawn of the 32-bit era and we were trying to figure out what we were going to do with all that newfound power. We hit a wall in the mid 1990s and there was precious little innovation for about a decade. I think we’re in another golden era right now, and there’s no reason to let incumbent-driven fear ruin it.

One thought on “SSDs, data loss, electricity, and hype

Comments are closed.

%d bloggers like this:
WordPress Appliance - Powered by TurnKey Linux