Backups — bad strategies

One thing’s for sure: until you have a backup strategy of some kind, you’re screwed, you just don’t know it yet.
— Jeff Atwood, What's Your Backup Strategy?, Coding Horror, 2008

So it turns out computers weren’t just a phase, they’re here to stay, and more and more of the sum of our lives is made, transmitted, or remembered as 1′s and 0′s in these rocks we’ve tricked into thinking for us.

So what happens when the music stops?

If you don’t have anything digital you care about, good for you, go and touch grass, skip this one. If you do have things you care about and you don’t have a robust backup strategy then you’re heading for a bad day. Your computer will die. Your phone will die. You’ll lose access to your email account. And it will all happen at the worst possible time. Or it won’t, but is it worth the risk?

Good backups allow you to recover from all reasonable risks. They are simple, routine, robust to seen and unseen loss, and tested and familiar in the event that they’re needed. If you’re backup strategy fails any of these tests then it is a problem waiting to happen. If it isn’t simple it will fail when you need it most. If it isn’t routine you won’t do it. If it isn’t off-site it’s not a backup. If it isn’t tested it isn’t real.

Examples of bad backup strategies

Having a single backup on an external drive that sits right next to your computer — or anywhere in the same house. It seems safe, until both are stolen in a break-in or destroyed in the same fire or flood.

Everything synced to Dropbox/iCloud/Google Drive or similar. If this is you, give yourself a brief pat on that back because you’re doing better than most, but then stop that and listen. This is not a good backup — synchronisation is not the same as backup. Two-way synchronisation is a surprising and insidious point-of-failure wherein damage (corrupted files, ransomware, accidental deletions) will silently propagate between synchronised devices, removing opportunities for recovery. By all means synchronise your files between devices, but don’t mistake this mechanism of convenience for a backup.

Having a routine system of backup that gives you peace of mind, only to discover that the backups had stopped running months ago, or some new folders weren’t being included. Untested backups are just wishful thinking.

Your backups are up to date and secured in multiple physical locations but you can’t access them because you’ve lost/forgotten the long pseudo-random password you encrypted them with — perhaps it was stored on your computer and your phone and you never anticipated losing both in the same incident. Having a complete backup on a hard-drive in your hand is useless if it’s encrypted with a key you don’t have.

You have backups, but no version history, and now you really wish you had the previous version of that draft of chapter 2 of your secret memoir, because you’re convinced you really preferred that version, but it’s gone and now you’ll never be the best selling author your year 6 English teacher always said you would be.

From those examples we arrive at a working list of scenarios we might want to guard against, which I’ll group according the kind of vulnerability they represent, and how each is best solved for.

Taken together, good backups are immutable, versioned, stored in multiple locations, with a known and tested recovery story that covers the loss of all your devices at once, no matter how unlikely that seems to you.

The universe tends toward maximum irony. Don’t push it.
— Jamie Zawinski, PSA: backups, 2007

In the next part of this series, Backups — a worst case scenario, I detail a specific worst case scenario which we will use to test our final backup strategy.