Little's Law Calculator
Little's Law: L = λ × W
Concurrency equals throughput times latency. Fill any two and the third follows.
requests in flight
arrival rate
time in system
Reading: at 100 req/s with 0.25s in the system, you need capacity for 25 concurrent requests. Size thread pools, connections and worker counts from this number, with headroom.
Request rate across windows
The same load quoted per second, minute, hour, day and month (30.44 days).
The one-line law behind most capacity planning, solved for whichever variable you are missing.
Little's Law FAQs
What is Little's Law?
L = lambda x W: the average number of requests in a system equals their arrival rate multiplied by the average time each spends inside. It holds for any stable system regardless of arrival patterns or queueing discipline, which is what makes it so useful.
How do I use it to size a thread pool or connection pool?
Multiply expected throughput by expected latency. At 200 requests per second with 150ms in the system, you average 30 requests in flight, so a pool of roughly 30 is the floor. Add headroom for bursts and latency spikes, because at the moment latency doubles, so does the concurrency you need.
What does it predict when latency degrades?
Trouble compounding. If throughput stays constant and latency rises, in-flight work rises proportionally, which consumes more workers and memory, which often raises latency further. Many cascading failures are this loop; bounded queues and load shedding are the brakes.
Which units should I use?
Any, as long as they agree. The calculator uses requests per second and seconds, so concurrency comes out dimensionless. Milliseconds work too if you divide by 1000 first.
Is anything sent to a server?
No. The arithmetic runs in your browser and the page works offline.