Building a Kubernetes Experimentation Platform with OpenFeature; Enabling Organisations to Put Theory Into Practice

The presentation discusses the use of feature flags in Kubernetes for experimentation and business value.

• Using feature flags as a mechanism for switching services to certain states simplifies blending features across a variety of services and enables changing inputs across services.
• Feature flagging allows for intentional experimentation to prove a hypothesis, rather than just testing infrastructure or code.
• Open Feature Operator is an example of how feature flags can be implemented in Kubernetes.
• Feature flagging can be used for conditional routing and proportional waiting.
• Centralized feature flag configuration can be used as an orchestra conductor for microservices.
• The presenter built an operator for feature flags as a real-world example.

The presenter used the example of a web server like nginx that can be extended with feature flags. Targeting keys can be used to respond to specific values, allowing for conditional routing. The feature flag configuration object can become a state for the configuration of microservices and the underlying cluster.

OpenFeature is a CNCF sandbox project that brings feature flagging everywhere. When combining the open-feature operator and Kubernetes, we gain the ability to add feature flags inside your workloads, controllers and operators. This creates a new opportunity for organisations to test infrastructure, create context-based feature gating and run large-scale experiments throughout the stack. In this talk, we will look at how feature flagging can be embedded and utilised to gain a level of declarative decision-making that enables cluster operators and engineers the ability to do everything from creating situation networking to managed failure modes. With live demonstrations and step-by-step walkthroughs of real-world implementations, the speaker will bring this project to life. There will be a special emphasis on how this enables experiments and how to act upon the results of those findings to both increase resilience and reduce cost.