Mastering API Architecture

Chapter 5. Deploying and Releasing APIs

In this chapter we will start to tie together how to move from design, build, and test to running in the target environment.

Consider the conference system case study we introduced in the Introduction: we had a single-user interface and server-side application. Deploying an upgrade to the server or user interface would likely mean having some element of downtime. It is likely that the deployment and the release actions are tightly coupled and possibly inseparable. It may also have taken time to roll back the changes if an issue occurred with the deployment. We will explore some options for the legacy conference system, in addition to looking at how a looser coupling between the UI and server components provides more options for deployment and release.

The introduction of traffic management provides you options to separate the deployment and release. In this chapter we will explore this in more detail and look at the conference system options available for rolling out changes. You will need to consider how API versioning impacts the options for modeling releases in the conference system.

One key consideration for a rollout is to understand whether a change has been successful or not. API architectures are by nature decoupled, and it is important to ensure the right metrics, logs, and traces are available to perform a successful release. We will look at the types of considerations for metrics and how they help in releases and incident management/troubleshooting. ...