Immutable Infrastructure Calls for Immutable Architecture

Abstract

With the advent of cloud computing and the concept of immutable infrastructure, the scaling and deployment of applications has become significantly easier. This increases the possibility of “configuration drift” as an operations team manages this cluster of machines, both virtual and actual. In this paper we propose a revised view on configuration and architecture. We propose that software deployed on a public or private cloud should, to the furthest possible extent, be immutable and source controlled. This reduces configuration drift and ensures no configuration problems in production as a result of updates or changes. We will show an example of a software project deployed on Amazon Web Services with an immutable Jenkins setup which manages updating the whole cluster and is self-regenerating. We will also discuss how this lends itself naturally to interoperability between clouds, because of the infrastructure-agnostic nature of this approach.