Back in June last year I wrote about creating a WordPress cluster on Amazon’s EC2. In this post I’ll run through a couple of the problems with that cluster, I’ve experienced, and how I solved them with Amazon’s Auto Scaling service.
The problems with the cluster
A couple of things were not ideal with the cluster that I’ve been putting up with for far too long but finally set aside some time this afternoon to fix.
1) The price of micro-sized spot instances spikes, to crazy prices, sometimes.
This has meant that although the price of micro instances while they’re running is cheap, when the price spikes they all die off and leave the cluster vulnerable. Unfortunately to set up auto-scaling in combination with spot-priced micro instances would require coding up a hybrid solution with shell scripts, and although I enjoy tinkering with this setup, I can’t justify that much effort when an out-of-the-box solution exists, Auto Scaling.