Moving fast in terms of development workflows aims to enable fast feedback by fast turnarounds. In order to increase productivity, developers who work on the application's behavior need to verify the implemented features and bug fixes in a timely manner. This includes the time spent on builds, software tests, and deployments.
The key to productive workflows is automation. Software engineers should spend as much time as possible on designing, implementing, and discussing business logic and as little as possible on cross-cutting concerns and repetitive tasks. Computers are designed for quickly and reliably performing deterministic, straightforward tasks. Humans, however, are better at designing, thinking, and brainstorming creative and complex tasks. Simple, straightforward processes that don't require a lot of decision-making should therefore be performed by software.
Build systems are a good start. They automate compiling, resolving dependencies, and packaging of software projects. Continuous Integration servers take this approach further. They orchestrate the whole development workflow from building artifacts to automated testing and deployments. Continuous Integration servers are the golden source of truth of software delivery. They continuously integrate the work of all developers in a central place, making sure the project is in a shippable state.
Continuous Delivery continues the approach of Continuous Integration by automatically shipping the built software to certain environments on each build. Since software changes have to be verified properly before they go to production, applications are first deployed to test and staging environments. All deployment actions have to make sure the environment is prepared and configured and rolled out properly. Automated and manual end-to-end tests make sure the software works as expected. Deployment to production is then done in a half-automated way by triggering the automated deployment manually.
The difference between Continuous Delivery and Continuous Deployment is that the latter automatically deploys each committed software version to production, if the quality requirements are met, of course.
All these approaches minimize the developer intervention required, minimize turnaround times, and improve productivity.
Ideally, the Continuous Delivery approach supports not only rollouts but also reliable rollbacks. Software versions, although verified before, sometimes need to be rolled back for some reason. In such a situation, there is either the way of rolling forward, for example, by committing a new version that will undo the recent changes, or by rolling back to the working state.
As mentioned earlier, software should be built in a reliable way. All versions of used technology, such as build dependencies or application servers, are specified explicitly. Rebuilt applications and containers produce the same result. In the same way, pipeline steps of development workflows should result in the same outcome. It is crucial that the same application artifact that has been verified in test environments is deployed to production later on. Later in this chapter, we cover how to achieve reproducible, repeatable, and independent builds.
In terms of reliability, automated processes are an important aspect as well. Especially, deployments that are executed by software rather than human intervention are far less prone to error. All necessary pipeline steps are well defined and implicitly verified each and every time they are executed. This builds confidence into the automated processes, ultimately more than executing processes manually.
Verification and testing are important prerequisites of Continuous Delivery. Experience shows that the vast majority of software tests can be executed in an automated way. The next chapter will cover this topic in depth. Besides testing, quality assurance also covers the software quality of the project in regard to architecture and code quality.
Continuous Delivery workflows include all steps necessary in order to build, test, ship, and deploy software in a productive and automated way. Let's see how to build effective workflows.