Simulate Your Cloud Architecture with an MCP Server Before You Deploy
Most capacity questions I get asked boil down to "what happens if traffic triples?" Usually the honest answer is "we'll find out in production." I build Cloud World Model (CWM), a simulator you can call over MCP, so an AI assistant can work through that question before deploying or changing infrastructure. This post walks through one real run: connecting, describing a small AWS stack, running it…
This article demonstrates how to simulate your cloud architecture using a Cloud World Model (CWM) simulator before deploying or making changes to your infrastructure. By connecting to the CWM server via MCP, you can run keyless simulations, which are temporary and have limitations on resources and steps.
To describe your desired stack, such as an internal ALB, two m5.large Node app servers, and a db.r5.large MySQL database running on max_connections of 500, you can use the simulation.create tool. The simulated environment uses 200 RPS as a baseline, with five steps. The simulation returned metrics like p50/p95/p99 latency of 4.52/15.01/87.10 ms, and an error rate of 0%.
The app CPU usage for each node was around 10%, while MySQL CPU usage was about 18.6%. Modeled DB connections were 40 out of a maximum of 500, and the estimated cost was $0.76 per hour.
Next, the simulation spiked the traffic to 600 RPS, and observed the system's performance under this increased load.
Written by urgent.news from Dev.to's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
