Make architecture trade-offs visible.

The problem

An architecture diagram shows components, but rarely helps its reader explore what changes under load or failure. This prototype connects a system drawing to an executable model.

Choices in the implementation

• Represent system components as typed nodes with constrained connections.
• Run simulation work in a Web Worker, with a main-thread fallback when worker startup fails.
• Persist the graph and simulation configuration so a design can be revisited.

What exists

• A React Flow canvas, preset scenarios and saved-design APIs with authentication.
• Queue, cache and topology models, adjustable request load, and node failure/recovery controls.

The next question

Compare the model with a small instrumented system and test whether users can explain a bottleneck after using the canvas.