At Simple, I worked on the operations team. I built and deployed infrastructure for the analytics team, which used a stack consisting of Kafka, Storm, and Amazon Red Shift.
Initially, we used Chef to provision the infrastructure, but experience with the software illustrated the benefits of immutable infrastructure, meaning that we didn't want to make changes to running systems. To that end I worked on Packer, fixing bugs and molding it to our needs. Our process then became building a new AMI every time we needed to make a change. This let us test the changes before deploying them, in a way that accurately mirrored how we would then deploy the images to production.
