Kubernetes API Challenges

Kubernetes is a powerful, flexible, container orchestration system, allowing teams to extend and shape it to meet their needs. It’s widely adopted across the industry and has a vibrant community of tools and services. For many organisations, providing Kubernetes as the de facto platform is the reality. That means teams get access to either a dedicated Kubernetes cluster or to a namespace within an existing cluster. Teams are free to deploy the services they need on their Kubernetes cluster. This freedom, however, comes with a price:

• High cognitive load for application teams. Kubernetes is a complex system, and there is a significant learning curve to getting started and managing. Developers need to invest time and effort to learn the platform and how to use it effectively.

• Waste of organisation time. A side-effect of the above is the amount of collective organisation time is invested in learning the technology. Multiple teams will go through similar journeys, and a lot of time may be invested in solving problems that other teams have already faced. 

• Lack of standardisation. While Kubernetes provides a powerful platform for building and deploying applications, there is a general lack of standardisation in how people use it across the industry. It’s expected that teams within the organisation will come up with their own patterns and standards. Aside from “reinventing the wheel”, this can lead to compatibility issues and make it difficult to collaborate with other teams.

• Monitoring and traceability. Part of the responsibilities of platform teams may include ensuring that the software running on the platform meets certain standards and requirements. As more and more teams use the platform, it may become increasingly challenging to determine if everything that’s currently running should be running.

• Does not raise the value line. Similar to “Cloud as the Platform”, providing raw Kubernetes does not raise the value line of the platform. Application developers are left to build the experience they desire from the Kubernetes primitives, while still having to take other aspects such as compliance, security and governance into account.

App teams have to manually discover and deploy the services they need