Ex Die in Diem


Data Redundancy

I only recently discovered that Dropbox is hosted on Amazon’s S3. This was a bit of a surprise, and not just because I use Amazon’s service to host my own website: it also had an impact on my data redundancy.

Being a bit paranoid, and a bit obsessive about little things, and very technically minded, I always have backups of my backups. To be more precise, for the things that matter to me I have at the very least two backups. For things that I really don’t want to lose, I have up to four separate backups. Just to be clear, that’s in addition to the working copy.

Let’s take an example. Like many people, I think my opinions and way of looking at the world are important enough that I’m writing a book. I know: you can thank me later. I’ve already put a fair bit of work into it, and so I’d be upset if I lost what I’ve written so far. In order to prevent such a loss from occuring, I back up the files that contain that work.

As well as being present in two different apps on my iPad, which is good if I accidentally delete anything, Both of those are backed up on a nightly basis to iCloud along with the rest of the system. I have had two opportunities to test this backup system on my iPhone, and I’d like to say for the record that it was seamless.

As well as the automatic backup, whenever I do a significant amount of work with these files I transfer them to their own little folder in my Dropbox. I also periodically update the zipped version of this folder that lives on Cloud, and make sure that it’s transferred to my mcnicholl.backup folder on S3. Four offsite backups, two local copies. And the paranoia simmers down.

Only, hang a second… didn’t I start this past saying something about Dropbx being hosted on S3? Oh no. Oh no no no. I was fine when I discovered that Cloud was run by Heroku and stored on S3, because that was just one link in the chain. Now it turns out that if S3 disappears, so does my backup there, and the one on Dropbox, and the Cloud version. And I’d be down to two local, one offsite.

(Quick pause to wind that tension back down. I’m saying here that if S3, owned and run by Amazon with all of their experience of web infrastructure, were to completely disappear at the exact moment that both of my local copies imploded, I could still just restore from my iCloud backup and be back to normal right away.)

Any normal person at this point would have realised that having an enormously redundantly distributed backup on some of the most dependable web infrastructure around as well as a further offsite backup and a local backup is probably a reasonable situation. Apparently I’m not a normal person, because yesterday I set up a webhosting account at cantab.net to bring my redundancy back up to four completely separate systems. What a waste of time and money.

This is from the