Infrastructure as Code, 2nd Edition

Chapter 14. Building Clusters as Code

Chapter 3 described an application hosting cluster as a service that dynamically deploys and runs application instances across a pool of servers (see “Compute Resources”). Examples of application cluster systems include Kubernetes, AWS ECS, HashiCorp Nomad, Mesos, and Pivotal Diego. This model separates the concerns of orchestrating applications from the concerns of provisioning and configuring the servers they run on (see Figure 14-1).

Figure 14-1. An application cluster creates a layer between infrastructure resources and the applications running on them

You should, of course, use code to define and manage your application cluster. As explained in Chapter 10, you can build an application cluster as one or more application stacks as code. In this chapter, I give some examples of what this can look like. I show some different stack topologies, including a single stack for a whole cluster, and breaking clusters across multiple stacks (see “Stack Topologies for Application Clusters”). I include diagrams for pipelines to deliver infrastructure code changes for these topologies.

I also discuss strategies for sharing clusters, or not sharing them, across environments and teams (see “Sharing Strategies for Application Clusters”). I close the chapter with a discussion of infrastructure for serverless applications (see “Infrastructure for FaaS Serverless” ...