Little's law on a closed loop: each virtual user sends, waits for the full response, thinks, then sends again. It assumes latency holds steady as load grows. Real services queue, so measured throughput bends below this.
Estimated throughput
45req / s
Per minute
2,679
Per hour
160,714
Peak (2× burst)
89 req/s
Egress
1.8 Mbps
Per day
3,857,143
Throughput vs response timeat 50 VUs
Every extra millisecond of latency costs you requests per second.A sensible starting point. Compare a measured run against 45 req/s. The gap is queueing you cannot see from here.
Estimates end here. Postman measures the rest.A run also has a shape. Pick one of Postman's four load profiles to see how virtual users are distributed across your 10-minute test.
Load profile
Fixed
peak 50 VUs
50 virtual users run for 10 minutes, each executing all requests sequentially.
What the maths assumes
One virtual user is one closed loop: send, wait for the full response, think, send again. It assumes latency stays flat as load grows. Real services queue, so measured throughput bends away from this estimate well before the numbers do.
Then measure it in Postman
Performance and Load testing runs a collection you already have at these virtual users and reports throughput, 90th-percentile response time and error rate live, broken down per request, so you can see which endpoint bent first.