The Shift to Declarative Continuous Deployment
November 21, 2022

Stephen Atwell

Imperative versus declarative deployment — which one is better? The answer: it depends. Development teams may prefer one over the other based on their workload capacity, infrastructure and desire for control. However, we are on the precipice of change. In the coming years, we will see a rapid shift to declarative automation in deployment.

Companies and development teams are under immense pressure to accelerate the software development cycle and improve the product while simultaneously controlling or reducing costs. More than 90% of developers and marketers say waiting to deliver improvements significantly impacts their business. Companies want developers and engineers to focus on the tasks that provide value — creating a stable program, fixing application bugs and delivering new features. After all, software delivery alone does not predict business success. It requires product reliability.

Let's look at why many more teams are gravitating toward declarative deployment.

What Are Imperative and Declarative Deployments?

Putting it simply, a declarative system says, "This is what I want to see in the end," and an imperative system says, "These are the steps I want the system to complete to get the desired results."

Imperative deployment is the more widely used strategy. DevOps teams build curated and explicit steps to take the program through continuous integration and continuous deployment, manually defining pipelines and the individual processes within. This paradigm is ideal for elite teams that want tight control over the process while keeping flexibility and customization.

Conversely, in declarative deployment, developers set deployment objectives based on a selection of variables. This approach focuses on the end result, not the steps in committing and releasing. Rather than telling the system how to deploy to an environment or what order to run tests, developers specify the code’s location, what environments it needs to reach and the requirements for each environment to receive the code. Modeling the process this way defines the desired application state, which provides the basis for the deployment and management system to automatically generate a deployment process that is more resilient to change than an imperative coded process. The system executes the deployment logic without involving the user. Developers can go back to writing code. This method is ideal for pre-elite teams who are focused on their product and care less about the deployment journey.

The Move to Declarative Technology Automation

The move from imperative to declarative in technology automation started outside of deployments. I previously worked for a different technology automation company that sold an imperative engine for automating service runbooks. Our most frequent automation happened when monitoring detected a service was down. The product would remove it from the load balancer, restart it and add it back. This was a huge time saver for many companies because it was typically done by hand in 2006.

Fast forward to today, Kubernetes is the declarative version. Instead of having a custom script triggered by monitoring, updating my load balancers and resetting servers, I simply state what workload I want to run and how many copies should be running. Kubernetes automatically monitors the health of the workload and fixes any detected issues. Once I declare a given state, Kubernetes is smart enough to maintain that state. I no longer need to tell my custom automation how to stay in that state. This is the difference between imperative and declarative.

The Move to Declarative Continuous Deployment

DevOps professionals began practicing imperative deployment before AI, when automation was not an option. The existing tool stacks have worked well but now limit how businesses scale their operations.

Developers are constantly creating new and better ways to develop software. This innovation has exponentially increased complexity. Developers face endless step combination options for the process. Because of this, they've got plenty to think about without needing to learn deployment skills and define the entire procedure. In fact, it's not a sound investment to train developers to become deployment experts — they should be coding. Developers just want deployments to work; they don't care about the underlying steps. Therein lies the benefit of the declarative paradigm — it simplifies automation, makes it more resilient to change, and eliminates manual tasks.

By removing the need for manual deployment supervision, developers have more time to write high-quality code. They benefit from the continuous feedback loop created by automatic deployment, which quickly informs them of issues and customer feedback and enables immediate remediation without diminishing customers’ service. Nearly a third of US consumers say they will leave a brand after only one bad experience, so releasing and maintaining reliable software is vital to business success.

Another benefit of declarative deployments: standardization. DevOps teams require advanced capabilities to ensure quality and stability, but many don't have the infrastructure or staffing to execute the same intricacies repeatedly. Automating continuous deployment ensures every change follows the same validation logic, simplifies delivery and makes it reliable, predictable and repeatable. Declarative deployment maintains uptime and regular operations.

Developers want to code, and that's what businesses need. Companies aren't going to increase value when their teams spend significant time pushing out updates instead of creating them. They need bug fixes and new features to drive customer satisfaction, skills only developers can supply. Some operations may stick with imperative deployment, and there's nothing wrong with that if it's working for them. The vast majority will see extreme advantages to adopting declarative practices. The technology exists, so why not use it?

Stephen Atwell is Principal Product Manager at
Share this

Industry News

March 20, 2023

To meet the growing demand for Oracle Container Engine for Kubernetes (OKE) with global organizations, Oracle Cloud Infrastructure (OCI) is introducing new capabilities that can boost the reliability and efficiency of large-scale Kubernetes environments while simplifying operations and reducing costs.

March 20, 2023

Perforce Software joined the Amazon Web Services (AWS) Independent Software Vendor (ISV) Accelerate Program and listed its free Enhanced Studio Pack (ESP) in AWS Marketplace.

March 20, 2023

Aembit, an identity platform that lets DevOps and Security teams discover, manage, enforce, and audit access between federated workloads, announced its official launch alongside $16.6M in seed financing from cybersecurity specialist investors Ballistic Ventures and Ten Eleven Ventures.

March 16, 2023

Hyland released Alfresco Content Services 7.0 – a cloud-native content services platform, optimized for content model flexibility and performance at scale.

March 16, 2023

CAST AI has announced the closing of a $20M investment round.

March 15, 2023

Check Point® Software Technologies introduced Infinity Global Services, an all-encompassing security solution that will empower organizations of all sizes to fortify their systems, from cloud to network to endpoint.

March 15, 2023

OpsCruise's Kubernetes and Cloud Service observability platform is certified to run on the Red Hat OpenShift Kubernetes platform.

March 14, 2023 released an update to the platform, delivering productivity for data teams.

March 14, 2023

CoreStack and Zensar announced a strategic global partnership. CoreStack will provide its AI-powered NextGen cloud governance and FinOps capabilities, complementing Zensar’s composable cloud operations offering.

March 14, 2023

Delinea introduced the Delinea Platform, a cloud-native foundation for Delinea's PAM solutions that empowers end-to-end visibility, dynamic privilege controls, and adaptive security.

March 13, 2023

Sysdig announced a new foundation that will serve as the long-term custodian of the Wireshark open source project.

March 13, 2023

Talend announced the latest update to Talend Data Fabric, its end-to-end platform for data discovery, transformation, governance, and sharing.

March 13, 2023

Descope has raised $53M in seed funding and emerged from stealth to launch a frictionless, secure, and developer-friendly authentication and user management platform.

March 09, 2023

Loft Labs announced Loft v3 with new capabilities and flexibility for platform teams to build and enable their development teams with a self-service Kubernetes.

March 09, 2023

AWS Application Composer is now generally available.