Throughput
Summary: The rate at which a system processes data or requests over a given period, serving as the primary performance metric for batch systems.
Sources: raw/chapter10
Last updated: 2026-04-18
In contrast to latency (how long a single request takes), throughput measures the total volume of work completed. For batch-processing, throughput is the total time it takes to crunch through an input dataset of a certain size (source: chapter10, p. 390).
Importance in Batch Systems
Batch systems are often designed to maximize throughput at the expense of latency. For example, MapReduce job execution is optimized for sequential I/O on disks and network efficiency, even if an individual task takes several minutes to start up (source: chapter10, p. 393).