About Load Balancer Simulator

Simulate load balancer distribution algorithms (round-robin, weighted, consistent hashing). Part of the DevTools Surf developer suite. Browse more tools in the Developer Utilities collection.

Use Cases

• Compare round-robin, weighted round-robin, and least-connections algorithms for a heterogeneous server pool.
• Simulate session affinity (sticky sessions) to understand performance trade-offs vs. pure load balancing.
• Model a consistent hashing ring to plan for node additions without cache invalidation spikes.
• Test health-check failure behavior to verify that traffic correctly reroutes around unhealthy nodes.

Tips

• Compare round-robin vs. least-connections under unequal server capacity — least-connections outperforms round-robin when servers have different response times.
• Simulate consistent hashing to understand how adding or removing a server node affects cache hit rates and request routing.
• Add health check failures to test how each algorithm handles node removal mid-session.

Fun Facts

• Round-robin load balancing was first implemented in hardware network switches in the late 1980s. Software load balancing at the application layer (Layer 7) became practical with Linux Virtual Server (LVS) in 1998.
• Consistent hashing was invented by Karger et al. at MIT in 1997 specifically for distributed web caching. It minimizes rehashing when nodes are added or removed — only K/N keys need to move (K = total keys, N = nodes).
• NGINX, released in 2004, displaced Apache as the most widely-used web server/load balancer by 2019, reaching 33% market share. Its event-driven architecture handles 10x more concurrent connections than Apache's process-per-request model.

FAQ

Which algorithms does it simulate?

Round-robin, weighted round-robin, least-connections, IP hash, random, and consistent hashing. Each shows request distribution across simulated server nodes with configurable weights and capacities.

What's the difference between L4 and L7 load balancing?

L4 (transport layer) routes based on IP/port — fast but can't inspect content. L7 (application layer) routes based on HTTP headers, URL, or cookies — slower but enables content-based routing.

Does it model health checks?

Yes — mark individual servers as unhealthy mid-simulation to see how each algorithm redistributes traffic. The simulator shows both active connections and in-flight requests during failover.