How to generate an AI Kubenetes diagram

Step-by-step guide on generating a Kubernetes diagram

1. Write a prompt describing the Kubernetes architecture. Applying the below framework generally yields good results:
   1. First, write out the structural components of the diagram. List out the pods, persistent volumes, ingress, and any other common components (https://github.com/kubernetes/community/tree/master/icons). Also list out the hierarchical groupings such as Kubernetes groups (e.g. clusters and namespaces) and network boundaries (e.g. regions, zones, VPCs).
   2. Then, write out how those components should be connected. An easy way to do this is to write out the sequential flow of how data or a request travels through the system (e.g. User sends request to the API endpoint, which gets forwarded to microservices).
2. Generate a diagram with the completed prompt.
3. Edit the diagram with follow-up prompts (this step requires signing in to Eraser).
4. Manually adjust the layout using GUI controls (this step requires signing into Eraser).

Tips on generating an Azure diagram

• Instead of writing a prompt from scratch, consider pasting excerpts from existing documents, call transcripts, code (e.g. Terraform, Pulumi, Azure Bicep), or uploading image files.
• Working with an LLM to create a fleshed out diagram prompt can be effective. Ask it to list out structural components as well as steps of a data or request flow through the system.

Example diagrams

• Baseline architecture for Azure Kubernetes Service (AKS)
• Distributed load testing with Google Kubernetes Engine
• Kubernetes cluster architecture