Bit Rot

[This week’s Communications column for the Vanuatu Independent.]

It’s happened again.

The life of a technology professional can sometimes feel like that of a doctor. You’re introduced to someone, and the moment you tell them your metier, their eyes take on a particular look and they say, “You know, I’ve been having this problem recently….” Immediately, the conversation becomes a diagnostic session.

I suppose everyone a lawyer meets has a court case pending, too.

So there I am, sitting down at a local cafe with a book and my morning coffee, and someone collars me with a request. “You need to write about CD Rot,” She says with a wry, knowing smile. Immediately, I put the book away. This is going to take some time.

(Before I go on, let me say that I actually enjoy these little conversations. If I didn’t, I would never have lasted as long as I have in IT. Heaven knows I wouldn’t be writing this column.)

CDs do rot, it’s true. Well, not ‘rot’ in the sense of wood decomposing. But they do decay. Everything does. Floppy disks used to be like mayflies in this climate, often not surviving the day. USB sticks are a little better, but after a given number of write operations (admittedly large, these days) they do have a tendency to go Poof. DVDs are no better.

Internal hard disks are a little more durable, by virtue of the fact that they tend hide inside the computer case, largely undisturbed. Still, it’s not unusual for Father Time to catch up with them as well, in the form of the ‘Tick of Death’ – a rhythmic, metronomic tapping noise made by a faulty drive continually failing to read, resetting itself and failing again, ad infinitum.

To a technician, the Tick of Death is as depressing as a death rattle. It means we have to turn to the anxious client and ask that dreaded question: “Do you have backups of your files?

We know you don’t, but against all hope we ask.

Beyond the inevitable alphabet soup of acronyms, computer geeks have a lexicon of silly metaphors, folk tales and scraps of half-clever word play they use to encapsulate otherwise hard-to-comprehend aspects of the digital world. One of my personal favourites is the term ‘Bit Rot’.

Now, bits aren’t organic. They don’t rot, per se. In a well-protected environment, they can remain exactly as they were for decades, theoretically forever.

But storage media – the things we keep the bits in – are not so kindly served by time. They do degrade. And when they do, we are given a vivid reminder of just how ephemeral the bits and bytes of information that reside there really are.

People want to treat data like a thing. They assume that if you buy some storage, and you put your bits in it, it’s more or less like a safe deposit box – as long as the thing looks okay from the outside, there’s no problem. Just pop the key in when you want to, and – hey presto – the data springs out, pristine and ready.

Would that it were so.

Data is dangerously fragile and ephemeral. It’s a not-entirely-accidental collection of electrical charges that manage to emerge in some useful order… most of the time. But shift just a few of those bits around, or drop a couple on the floor, and the whole construct become no more intelligible than line noise on a telephone wire.

Larger institutions with a significant investment in the information they’ve stored often spend inordinate amounts of money ensuring that their bits remain ordered and accounted for. Often enough, this takes the form of large tape libraries or – more common today – vast, redundant collections of hard disks.

The silly ones invest heavily on a monolithic construct built out of top-quality equipment. While it may be safely said that you really do get what you pay for when it comes to computer equipment, paying top dollar for a single basket in which to store your bits is a mug’s game. Three straw baskets is vastly better than one silk purse.

Google is a little smarter in their approach. No one knows the exact amount, but recent informed estimates put the number of servers they operate at well over 100,000. Each of those machines consists of a fairly decent quality board with two processors, a modest amount of RAM and a hard disk. There’s no case, little duplication of core components, and it’s slapped bare onto a run of the mill computer rack.

When one of their servers dies, they send a technician to pull the plug, and leave it right where it is, until the rack itself is useless.

You see, the folks at Google are smart. They don’t really care about the computers. They care about the data. And they make sure that they always have another copy lying around. That’s not a completely trivial task; it’s why they have some of the best computer engineers in the world working for them.

But the premise they work from is remarkably simple: Make sure you have at least 3 copies of every important file at any given time. The first copy is the one you’re working on. The second copy is nearby, so you can swap it with the one you’ve got without losing any time. The third one is sitting some place safe, preferably very far away.

This is the fundamental principle that underlies just about everything in the formal field of Data Integrity. Sure, it’s a field that you can spend a career in and still not know enough. But until you learn this first, basic lesson, you’ll never learn anything: Data that doesn’t exist in at least two places… doesn’t really exist.

Make copies early and make them often. Before you change a file, make a copy. Once you’re done changing it, make a copy. When you store it, make a copy.

Store those copies well. Keep one of them at hand, say on your computer or laptop hard disk. Keep another nearby, on a USB stick or a CD. Then – and this is important – keep another one offline some place. Send it to your GMail account as an attachment (these guys know how to protect data, remember?), or work out a deal with a colleague or a friend to store files on their machine.

Keeping and managing multiple copies of your data is a simple process in principle. In practice, it’s a pain in the posterior. But unless you want to have a long, commiserating chat with me over coffee, I’d recommend you find a way.

But if you’ve only one copy of your files, I guarantee you it’ll end in tears. I promise to be sympathetic, and I’ll help you if I can, but if you haven’t taken those first steps to protect yourself from Bit Rot, there’s not a lot I can do. Except maybe pay for your coffee.


Addendum: In the course of dashing off this piee, I mistakenly over-write last week’s Communications column. Happily, I do occasionally take my own advice. A saved copy attached to a message in my GMail account allowed to reclaim the ‘lost’ article.

So-o… I told you so.

Now go back up your data.