Last weekend, I spent 48 hours rebuilding and moving all the backend infrastructure around. All that happened was a very short outage while we finished moving everything around between instances, but what this MEANS is that now things are faster, better, and more reliable!
If you’re not a nerd, feel free to ignore the rest of this post. However, for the nerds, this is what we’ve done:
- We now have a multi-master database cluster, with a load balancer and floating IP address to ensure that we have zero database downtime
- We’ve split out all our worlds into separate instances. This means if a server needs a reboot, we can reboot the entire instance that the world is on, without having any effect on any other worlds!
- Dedicated t2.medium instance that runs map rendering. This means we can start rendering the OTHER worlds on maps.mcau.org
- The ability to easily and automatically create and destroy servers and worlds on demand.
- Other cool nerdy things (ansible! redis! memcache!)
What this means for you is that things will be faster, better, and more reliable. What this means for US is that we’ll be able to do cooler things, easier, and faster!