Continuous Integration and Delivery

State Januar 2019

The automation of build, test and deployment automation has continued to be an important cornerstone of modern software engineering. Continuous integration as well as continuous delivery support development teams to become DevOps teams by sharing responsibility for all aspects of the product lifecycle.

The concepts of CI/CD have a rapidly increasing effect on the operations of software components. Infrastructure-as-code enables the management of infrastructure components in the same way as software components.

At the same time, CI/CD products have evolved to bring the concept of everything-as-code to the world of pipelines. Many products support pipelines to be expressed in code and have adopted YAML to be the language of choice. The advantages of pipeline-as-code include:

The textual representation of pipelines enables an easier understanding of the inner workings. If used with code folding, the pipeline can easily be explored on the required level of detail.
Pipelines usually closely matche the design of a software component. With pipeline-as-code, the pipeline can be stored in same repository as the corresponding code. This enables the pipeline instructions to evolve with the code and ensure that both work together.
The reuse of pipelines are greatly improved by handling text files instead of working through multiple layers of configuration dialogs to customize a cloned pipeline to the new use case.

Over the last years, containers have revolutionized the way software is developed and deployed. This also applies to CI/CD where container native builds isolate the steps of a pipeline in individual containers. This has enabled CI/CD to profit from the advantages promised by containers:

Pipeline jobs are executed isolated from the underlying system. This greatly simplifies the configuration of build environments
By using containers, the build environment comes in an immutable package - the container image. This can be created by the developer to suite the needs of the build and deployment process
The standardized execution environment allows for pipeline reuse

Revisions:

scale | November 2017

Why?

In former times the release cycle for new features in software applications was months. Planning, Conception, Implementation, Integration, Testing, Deployment were sequential tasks. Most of them were done manually. Integration tests are time consuming and brittle. Deployment was done by specialists with a lot of knowledge buried in their head.

Continuous Integration(CI) and Continuous Deployment(CD) targets improving that situation dramatically.

Continuous Integration (CI) is the practice of testing each change done to your code base automatically and as early as possible.

Continuous Deployment (CD) follows the testing that happens during Continuous Integration and pushes changes to a staging or production system. This makes sure a version of your code is accessible at all times.

Fast feature releases Customers get new features in days and not months.
Reduce integration problems Integrate new changes immediately and automatically check whether they work
Catch issues early Issues are fixed immediately
Highly automated processes that run with high-stablility No tacit knowledge anymore. Reliable, automated processes.
Reduced risk Depoyment in small steps instead of the big bang reduces significantly the risk of deployment.
More features Reliable processes give you more time to develop new features.

Automation is a cornerstone of a great development workflow. Every task that can be done by a machine should be. Automation gives you the time to focus on new features. But the implementation of automated test, integrations and deployments does not come for free.

Technology choices/solutions

We use a lot of different tools for the purposes of automated testing, automated integration and deployment. This is only a subset of our toolset.

Artifactory
GoCD
Jenkins
TFS Build system
Azure/AWS APIs and scripting

Our projects

This is also only a small representation of teams that have automated processes.

Foundational services Highly automated processes from unit testing and integration testing and deployment to Azure. The team puts a lot of effort into improving the automation and stability of feature releases.
iDesk
Lexware desktop apps Automated build processes, including setup creation, setup installation and basic tests.

Contact