service-time · shipping · also service requirement

Service time

Service time is the processing time required by a request at a specified resource, excluding time spent waiting for service there.

For one nonpreemptive stage, service time S is completion minus service start; queueing delay Q precedes service, and response time R = Q + S. With preemption, count actual service across its separate intervals. [1]

Asynchronous fio clat ends when fio reaps completion, so it is not an isolated device-processing measurement. Matching units do not make different timing boundaries equivalent. [2]

The existing synthetic FIFO trace has a request arriving at 10 ms, starting at 101 ms and completing at 102 ms: S = 1 ms, Q = 91 ms and R = 92 ms. The model supplies the service-start event; an observed completion alone does not. [3]

In a separate constructed preempt-resume trace, request A runs during [0, 2) and [6, 8) ms. Its 8 ms residence contains 4 ms of service and 4 ms of later waiting, despite starting immediately. Completion minus first start would overcount service. The example assumes one server and no overhead or other state. [3]

Relations

related
Queueing delay, Storage latency, Completion latency (fio), Little's law, Pollaczek-Khinchine mean formula, Kingman approximation, Tandem queue, Shortest remaining processing time, Request slowdown, Processor sharing

Citations

  1. Harchol-Balter and Scully: The most common queueing theory questions asked by computer systems practitioners, sections 1 and 2.1. (accessed 2026-11-18)
  2. fio manual: interpreting slat, clat and lat; asynchronous completion observation. (accessed 2026-11-18)
  3. Hesela: separating waiting and processing in the existing synthetic FIFO trace. (accessed 2026-11-18)

JSON: /data/terms/service-time.json

Published by Hesela team · Updated

Report a correction · Editorial standards · Reuse guidance

Cite this page
Hesela team. Service time. Hesela. 2026-11-20. https://hesela.dev/terms/service-time/