Beyond Block Diagrams: Different Ways of Understanding K8s Architecture


Authors:   Kim Schlesinger


Alternative ways of visualizing and understanding Kubernetes architecture for early career engineers with limited computing experience
  • Create diagrams that show the passage of time to represent Kubernetes clusters
  • Use distributed tracing tools as a learning tool to identify errors and understand the system
  • Build 3D models of Kubernetes clusters to help people develop a richer mental model of the system
The speaker, who is part of the cloud native generation, did not have prior experience with the technologies and practices that Kubernetes is replacing and found block diagrams difficult to understand. To help early career engineers with limited computing experience, the speaker presented three alternative ways of visualizing and understanding Kubernetes architecture: time sequence diagrams, combining distributed traces with a diagram, and 3D models of Kubernetes clusters. The speaker emphasized the importance of hiring, training, and retaining early career engineers and being deliberate about helping them understand complex systems.


Most introductory Kubernetes videos and blog posts use static block diagrams to illustrate the core components of Kubernetes and how they interact. These diagrams don’t convey the dynamic communication between the K8s control plane, nodes, and pods, and we owe it to ourselves and people new to the field more accurate and engaging ways of representing Kubernetes architecture. This talk will demonstrate a few different methods to visualize how Kubernetes components work together including a time-sequence diagram that models what happens when a user creates a pod using the Kubernetes command line tool, kubectl, and how to use Jaeger to see an http request move onto different nodes by traveling through a load balancer, to a service, and arriving at a specific pod. This talk is for people who are new to Kubernetes and want to understand how the system is designed, as well as experienced engineers who are curious about alternative ways of visualizing Kubernetes architecture.


Post a comment

Related work

Authors: Matt Broberg, Kohei Ota, Kaslin Fields, Kat Cosgrove

Authors: Michael McCune, Guy Templeton, David Morrison, Joachim Bartosik