Last year I wrote about learning every year and its importance to me.
In 2021, the Kubernetes learning journey continued. There is always a difference between getting something just working and getting it working well and I was aware that there were many large parts that worked but I didn't fully understand. The real test for any orchestration system is how it handles failure and so running multiple VMs across multiple physical hosts, with high availability configured, was a great way to exercise this. Deploying new services and apps to Kubernetes quickly builds more comfort with Docker too and that practice has been good. Once the container habit is ingrained, it's a really clean way of maintaining dependencies, and reducing a class of failures.
This year I added more complex storage into the mix, setting up my first Ceph cluster. Ceph is designed for much larger scale than I'm using it for, but has given a chance to understand the decentralized design, techniques like erasure encoding, and how to design a system for resilience.