Throughput

Turkish equivalent: İş hacmiDomain: Performance Engineering

The amount of useful work a system completes per unit time under a defined workload and correctness boundary.

Throughput is the amount of work, requests, records, or data a system successfully completes per unit time; it is not the same metric as latency.

Measurement and Bottleneck Boundary

Throughput is a rate of useful completed work, not a synonym for low latency or high CPU utilization.

Throughput Must Be Read Together with Latency

Higher throughput does not automatically mean a better system. Larger batches or greater concurrency can raise work per second while worsening P95/P99 through queueing. Capacity evaluation should measure operations per second, tail latency and saturation at the same load point.

Related technical publications

Publications whose title or summary directly references this concept.

High-Performance Java Data Systems

A comprehensive engineering note on latency, throughput, JVM behavior, JPA/Hibernate, JDBC, HikariCP, SQL, indexing, caching, Kafka, PostgreSQL, load testing, and capacity planning for high-performance Java data systems.