Continuous Delivery Pipelines

Continuous Delivery Pipelines

by Vijay Kumar

Transitioning to an agile enterprise makes sense for any business that hopes to remain competitive today. So of course, continuous delivery is a critical part of any organization. Not only does it enhance the accuracy and speed of software development at scale, but it also allows you to react to any market, technology and management changes that may occur during the course of software development.

What is Continuous Delivery?

Continuous Delivery is a software engineering approach that focuses on developing software in short cycles to ensure regular, frequent releases. While the traditional approach focuses on software development from start to finish (causing delayed releases), Continuous Delivery ensures the availability of stable, quality builds that are ready to be released when needed.

So, can this be improved further?

An Efficient Continuous Delivery Pipeline

The Continuous Delivery Pipeline attempts to automate several parts of this process and optimize it even further. Today, a lot of tools are available to to help agile organizations implement efficient CD pipelines quickly.

The pipeline essentially combines automated builds, testing, and deployment into a single release workflow. In a way, this creates and nurtures a Testing Driven Development environment that ensures speedy releases without compromises on quality.

A CD pipeline shouldn’t compromise on three critical pillars (even sacrificing one for the other is unacceptable in most cases):

  • Quality – even though speed is important, it should never come at the cost of quality. Deploying faulty code faster ends up doing more harm than good.
  • Frequency – a continuous delivery pipeline should push quality code frequently. This ensures regular MVP (Minimum Viable Product) assessment, allowing any changes in requirements to be addressed almost immediately.
  • Predictability – product deliveries should not just be frequent and of top quality, the delivery should be predictable as well. This helps unclog the DevOps process even more by removing any bottlenecks.

An efficient Continuous Delivery Pipeline isn’t just the fastest to get code to production. It takes quality code to production both frequently and reliably.

Stages of a Continuous Delivery (CD) Pipeline

Although there aren’t clear-cut rules or guidelines regarding CD Pipelines, there is a generally agreed-upon rudimentary structure that proves to be the most effective. Figure 1: An illustration of each process filtering the impurities and feeding back to previous stages.

The stages are split into 5:

Commit Stage (CI)

The first stage in the CD pipeline deals with the phase of code commits. All new code is usually housed in a self-contained file system and an automated series of tests tries to correct code and functional errors at an early stage. These tests are divided into two –

  1. Unit tests – Code is isolated into smaller parts, called units. These units’ individual methods and functions are then tested for any errors.
  2. Split or A/B Tests – Two versions of code are tested at the same time to find the version with better functionality. The better version is then sent on to the next phase.
Automated Testing/Acceptance

All the code that passes the commit phase is sent for automated testing. Here, the current version of code is tested in a near-production environment to make sure that all code passes basic quality standards.

Basic functionalities such as the pressing of buttons and launching and exiting an application are tested. Once basic functionality is successfully tested, the code is then stress tested to ensure stability.

Functional tests can include –

  1. Internationalization
  2. Localization
  3. Data Quality
  4. Accessibility
  5. Negative scenarios

Code that passes these tests is sent on, while any errors at this stage sends it back to the development team with feedback. The automated testing stage ensures that common flaws and failures are detected and fixed early in the development process.

Alpha Testing

Alpha deployments are used by developers to check the code changes made up to this point. Moreover, the exploratory nature of this testing relies on intuition rather than a fixed series of steps/tests.

Developers explore the code and see the changes they made first-hand. In addition to checking the newer code, developers also check whether they inadvertently broke something in the main branch of the application.

Newer code is checked for compatibility with older code; any errors are fixed immediately, and the code is sent back to the previous stage.

Beta Deployments

A beta testing phase should be manual. In this stage, testers (QAs) test the application to uncover missed bugs and errors.

Although a manual stage in a pipeline based around automation might seem counter-intuitive, it is a crucial step to ensure that only quality code gets to production. Moreover, the automated testing phases get rid of most issues before this stage – ensuring that manual testing doesn’t take too much time.

Production

The Continuous Delivery pipeline ends with the production stage. All code reaching this stage is validated to meet given specifications. In addition to meeting all requirements, the extensive testing guarantees that only quality code gets to this stage.

Of course, the actual release timeline for feature releases might depend on business and marketing aspects of the organization. Once a certain build is tested and production is complete, the pipeline gets ready for the next build.

Vijay is a Senior DevOps Engineer and Systems Architect at Practical Logix

Leave a Reply

Your email address will not be published. Required fields are marked *