What you are seeing
Particles are tasks. They arrive from the producer, wait in the queue, run through one worker, and disappear when complete.
Interactive Systems Explainers
Queue Congestion Playground
See why latency explodes when work piles up.
Key Observations
What to Notice
Why latency grows
When arrivals are only a little faster than processing, waiting work accumulates. Each new task must stand behind everything already stuck.
Throughput vs latency
The worker may keep finishing tasks at the same rate while latency gets worse. Output can look stable while the line behind it grows.
When queues start dropping work
Turn on the queue limit and overload the worker. Once the line fills, new tasks bounce away instead of waiting forever.
Real-World Context
Queues are useful shock absorbers until they become places where overload hides. They sit behind background workers, message brokers, request handlers, database writers, email systems, video processing jobs, and event pipelines.
The risk is deceptive calm: throughput can look acceptable while user delay, memory use, retries, and timeout risk get worse. Queue limits, admission control, and worker counts are choices about what should wait, what should fail fast, and how much delay the system can tolerate. Backlog age is often more revealing than queue length alone.
Interactive Systems Explainers