Wednesday, December 02, 2015

Ruminating on the 'Infrastructure as Code' paradigm

Setup (installation and configuration) of servers and other infrastructure components is a tedious process.

In the olden days, process-oriented teams created meticulous documentation on 'how to setup a server' - typically in a word document with screenshots and step-by-step instructions.
Folks then tried to automate some tasks using scripts - e.g. unix shell scripts/bash etc.

But today, in a cloud-first world, setup of servers and deployment of applications need to be completely automated. The whole premise of 'Infrastructure-as-Code' is to write code in a high level language (e.g. Java, Python, Ruby) or a DSL (domain specific language) to automate the provisioning of infrastructure and managing configurations.

So this goes beyond just writing simple scripts. You utilize all the best practices of agile development projects - i.e. version control, unit testing, iterative development, etc. The whole revolution happening in DevOps acted as a catalyst in promoting the concept of 'programmable infrastructure'. In DevOps, the core concept of 'You built it, You run it' promotes closer collaboration between the development teams and IT ops team.

Popular tools such as Ansible, Kubernetes, Puppet, Chef, etc. can be used to automate your complete deployment cycle and help you achieve Continuous Delivery.