tomtaylor.co.uk

Amazon S3 (Simple Storage Service) is an incredibly cheap way of making backups that will survive theft, fire or just a broken hard disk. You pay about $0.15/GB/month for the storage, $0.10/GB for data transfer and a few more cents depending on how many requests you make. I’m storing about 4Gb and I pay about 40p a month - stupid cheap.

Probably the most important things I have on my computer are my photos, especially the RAW originals. Since I’ve started using Aperture to process my photos, it makes sense to backup the library somewhere secure… like S3.

Aperture’s library looks like a single file, but it’s actually a bundle, comprised of lots of metadata and structure. You can have a play - right click on the file and select ‘Show Package Contents’. The upshot of this is that it’s very easy to synchronise the changes to Amazon S3, copying new files and changes without having to move the entire library every time.

In steps s3sync, a Ruby script modelled off rsync that will keep a remote copy of your files on S3 in sync with a local directory (or visa-versa). The README file explains the installation and setup. I’ll run over it if there’s a demand.

The command you want to run is:

./s3sync.rb -rv --exclude="AP.Minis|AP.Thumbnails|AP.Tinies" --progress --delete ~/Pictures/Aperture\ Library.aplibrary bucketname:backups/aperture

We exclude AP.Minis, AP.Thumbnails and AP.Tinies because these are quite large and Aperture regenerates them every time you make a change. They will be regenerated if they don’t exist. We use delete to remove anything that doesn’t exist locally anymore, keeping S3 properly in sync.

Remember, it takes a long time to perform the backup because you’re potentially uploading a massive chunk of data - but you can cut it off (Ctrl-C) and resume again later.