Throughput
Throughput is the number of requests or amount of data processed per unit time (QPS/RPS/BPS), measuring system capacity.
Purpose
Quantify system capacity, guiding capacity planning.
Mechanism
QPS = requests per second. Bounded by CPU/memory/IO/network/locks. Little's law: L = λW (average concurrency = arrival rate × average latency). Throughput and latency usually trade off: high throughput often raises latency.
Use cases
- Performance measurement
- Capacity planning
- System design