Back to Projects

Resilience Visualizer

An interactive browser-based simulator for microservices resilience patterns — circuit breakers, thread pool exhaustion, cascading failures, and more.

React TypeScript Vite Zustand React Flow Tailwind CSS Recharts Vercel Claude Code
Live Demo

About the project

Resilience Visualizer is an interactive simulator for microservices resilience patterns that runs entirely in the browser — no installation, no backend, no infrastructure.

The idea came from a recurring problem: concepts like circuit breaker, thread pool exhaustion, or cascading failure are hard to explain with words or a static diagram. You can draw arrows and boxes, but that doesn’t show what happens over time — how one slow service can bring down an entire system, or how a well-configured circuit breaker isolates the failure and protects everything else.

The tool lets you draw any microservices architecture as a graph, fire real-time traffic through it, and watch exactly what breaks and why.

Demo

Screenshots

Resilience Visualizer – system running normally
System running normally — nodes are green, requests flow through the BFF architecture with a load balancer.
Resilience Visualizer – cascading failure in progress
Cascading failure — an overloaded backend exhausts BFF threads, which starts rejecting client requests.
Resilience Visualizer – retry storm
Retry storm — a sudden traffic spike overwhelms the gateway, threads are exhausted, the system enters an error spiral.
Resilience Visualizer – e-commerce microservice mesh
Microservice mesh — Gateway → User/Product/Inventory Service → separate databases. A realistic production scenario.

Key features

  • Circuit Breaker with three threshold modes: Count, Percentage, Both — full closed→open→half-open→closed lifecycle
  • Platform vs Virtual/Async threads toggle — demonstrates how blocking I/O causes thread starvation vs async models
  • Load Balancer with round-robin, random, and least-connections strategies
  • Kill / Recover buttons per node for chaos engineering — configuration is preserved on recovery
  • Dual error metrics: windowed Err%(W) for fast node coloring, cumulative Err%(∑) for long-term trends
  • 8 built-in scenarios: Cascading Failure, Circuit Breaker Demo, Retry Storm, Microservice Mesh, Timeout Tuning, and more
  • Real-time metrics per node: RPS, avg/p99 latency, error rate, thread pool and connection pool utilization
  • Speed control from 0.1× to 10×

Try it

resilience-visualization.vercel.app →

Open it in your browser, pick a preset scenario, hit Play, then kill a node and watch what happens.

Read more

For a deep dive into the simulation engine, the bugs I found along the way, and how Claude Code helped build this — read the full article.