Capacity Planner
Work out how many workers and servers a service needs from peak requests per second and response time using Little's law, with headroom and memory per server.
Free · No sign-up · Your inputs stay in your browser
Input
Plan capacity with Little's law
Output
96.0 requests in flight on average
Workers needed
125
96.0 busy on average plus 30% headroom
Servers
8
16 workers each; add one more so a deploy or failure does not take you under
Worker memory per server
1.3 GB
Throughput ceiling
1,066 req/s
With every worker busy and latency unchanged.
Little's law: the average number in a system equals arrival rate × time in the system (L = λW). It needs no assumption about the traffic's distribution, which is why it is the first sum in any capacity plan. Size for the peak, not the daily average, and remember that latency rises as workers saturate: once requests queue, W grows and L grows with it.
Runs in this browser. Nothing you paste is sent to a server.
About this tool
Method, assumptions, practical next steps, and limits.
Little's law
L = λW holds for any stable system regardless of how requests arrive: 800 requests per second that each take 120 ms means 96 requests in flight on average. That is the minimum number of workers that are busy at once. The law is named after John Little, who proved it in 1961.
Why headroom
Averages hide bursts, and when every worker is busy new requests queue, their response time grows and so does L, a feedback loop that turns a busy period into an outage. Size for peak traffic with headroom, and keep one more server than the sum so a deploy or a failed node does not push the rest over.