Amazon Glacier is a cheap massive cloud storage solution that is mostly suitable for storing cold data - data that are rarely accessed. The price is fair: $4/TB/month. However, it's not like Dropbox or Googld Drive that has nice client programs that you can simply drag and drop the files to be stored. Instead, you'll have to work with their APIs to upload you files. In this post, I'll explain the basics about how to upload the files and also how to query the inventory.

Now that we can take advantage of rsync to minimize the data to transfer when backup. But it's still a little uncomfortable if we need to do this manually everyday, right? Well, cron is here to solve the pain.

Recently I do most of my work on a remote data center through a slow network connection (<100KB/sec). I usually backup my project source tree as follows. I first do make clean and also delete any unnecessary obj files to shrink the total file size, then I compress the whole source tree as a tar ball and then I use scp locally to fetch the backup tar ball to my local machine. The procedure is quite boring since I need to do this every day before I go home, otherwise the whole bandwidth will be occupied for near an hour during which I can almost do nothing.

Situation gets better when I find rsync and cron. Here is how I do automatic regular (daily) backup with them.