Round Robin
Round Robin sends each request to the next server in order. It looks tidy when servers are similar, but weaker servers can fall behind under pressure.
Interactive Systems Explainers
Load Balancer Playground
Watch routing choices spread traffic and create queues.
Key Observations
What to Notice
Random
Random routing scatters requests without looking at server load. Watch for uneven clusters and sudden local congestion.
Least Connections
Least Connections favors servers with less active and queued work. During spikes, it tends to spread stress and recover more smoothly.
Weighted
Weighted balancing gives stronger servers more traffic. The weak server receives fewer particles, so its queue usually grows more slowly.
Queues
Stacked particles are waiting requests. When arrivals outpace processing, the stack grows and the server starts glowing hotter.
Overload
Red glow, jitter, and slower motion mean the system is stressed. Congestion can push backward toward the load balancer.
Real-World Context
In production systems, load balancers are often where traffic policy becomes visible. They sit at the edge of web apps, in front of API fleets, between internal services, and beside caches or gateways.
Routing policy shapes failure isolation, noisy-neighbor behavior, deploy safety, and how quickly unhealthy instances stop receiving work. Simple rules are predictable, while adaptive rules react to pressure but can be harder to reason about. During incidents, small routing choices can decide whether users see localized slowdown or a wider outage.
Interactive Systems Explainers