Generic helm templates push variant drift management into values files, which are not amenable to reusable tooling.

Score is a developer-centric and platform-agnostic workload specification. It ensures consistent configurations between local and remote environments.

We can use AI to build deterministic, domain-specific tools that can inspect, modify, and validate the desired state based on stable APIs.

Transformation is a powerful technique used by config-as-data tools, such as kustomize, kpt, and ConfigHub, but it’s a means to an end.

Rolling out new configuration policies across an organization with configuration as code and policy as code tools can be a challenge.

Validating triggers provide the kind of guardrails that AI agents need in order to verify their work and guide their act-and-fix loops.

Is generating Kubernetes configuration with AI better than alternative methods of authoring the configuration?

The lack of well defined, discoverable application interfaces impedes automation and management of applications, and obstructs configuration validation and generation in particular. Ten years later,…

Can the evolution of operational tools over the past 4 decades provide clues about what changes AI will bring to ops?

We can build composable, interoperable tools to automate cross-cutting configuration changes on top of configuration as data.

Where does GitOps go if AI manages all our configuration? A proposal to extend the GitOps principles for config that is generated and processed at machine speed — call it Generative GitOps.

Configuration as data is a paradigm shift that enables platform teams to query and fix security policy across whole fleets of clusters.

Last time I built a container vulnerability scanner on ConfigHub by domain-swapping Brian Grant’s RBAC Manager. This post does it again…

Hello New World!

Configuration data is not just text, but is not just JSON, either. It requires some special approaches to effectively represent it as data.

“Configuration as Code” trades off API-driven automation for scripted variant creation. What have we traded away?

Configuration as code has a huge opportunity cost: the ability to create apps that change state. Build K8s apps using configuration as data.

The history behind Kubernetes managed fields, an explanation of what it is for, and introduction to a tool to manage them.

How Kustomize, Kpt, Porch, and ConfigHub are different, and why each one was created.

Introduces a Kubernetes package installer designed for simple installation and post-installation customization.

I used an AI agent to convert a Helm chart to an Agent skill, and discuss the benefits of using skills to install workloads on Kubernetes.

Kubernetes does not roll out changes to ConfigMaps automatically. This discusses the history of the issue and the two prevailing solutions.

Kyverno is a policy-as-code tool for Kubernetes that enforces state-based policy constraints, which can be used with ConfigHub.

What is Kubernetes Strategic Merge Patch, and why does it exist?

Abstraction of configuration through parameterization and encapsulation of configuration generators contribute to configuration complexity.

Startup ConfigHub treats configuration as data rather than relying on endless opaque files generated in CI/CD pipelines.

If configuration had an API, you could build and use tools to edit, inspect, and validate configuration.

An explanation of bidirectional GitOps and why and when it can be useful.

Variants — similar but different related configurations — are very common. How does ConfigHub manage them using WET configuration?

Why ConfigHub manages configuration as data.

A simple, concrete example of Configuration as Data.

Explanation of “Configuration as Data” and its potential benefits.

I don’t want helm charts to be my interface to off-the-shelf components on a daily basis. Kubernetes resources are simpler.

Coding agents can now find and cherrypick the context they need, and plan and perform multi-step tasks.

A glossary of Kubernetes Configuration and Infrastructure as Code (IaC) terminology.

What tools are available to manage applications across a fleet of Kubernetes clusters?

PearX is our accelerator for elite founders. The small batch program has a high success rate - over 90% of companies raise a seed round.

Cloud-native leaders launch ConfigHub to unify configuration management with modern automated development workflows and compliance.

What is Infrastructure From Code, how does it differ from IaC, and what are its benefits and challenges?

From the beginning of the Kubernetes project, I had ideas and opinions about how declarative configuration should be supported. That led to…

There is an another world of automation tools outside Infrastructure as Code (IaC) that can achieve many of the same objectives, and even…

Can configuration languages dramatically reduce configuration complexity?

Should all developers deploying to the cloud and/or Kubernetes learn how to use Infrastructure as Code (IaC) tools?

Infrastructure as Code tools currently have no awareness of the purposes of configured resources or of the operations being performed.

How Infrastructure as Code is different from imperative user interface surfaces from an implementation point of view.

If you’ve used Infrastructure as Code (IaC), you’ve almost certainly used IaC packages. This post digs deeper into their uses and pitfalls.

Variant generation, i.e. the generation of multiple distinct configurations from the same IaC code, is the primary capability of IaC.

Practices for building and deploying applications are not ideal for provisioning via Infrastructure as Code, due to critical differences.

I discuss twelve “anti-factors” that make Infrastructure as Code suboptimal. This recaps my IaC posts from 2024.