rate-latency-service-curve · shipping · also Rate-latency server

Rate-latency service curve

A rate-latency service curve gives a minimum-service guarantee with rate R and latency parameter T through beta(u) = R*max(0,u-T).

Guarantee: cumulative output D satisfies D(t) >= inf over 0 <= s <= t of A(s) + beta(t-s). This does not promise service while no work is available. R is a guaranteed share, not necessarily the physical transmission rate. [1]

Fluid bound: a FIFO lossless flow constrained by burst b and rate r <= R has delay at most T + b/R and backlog at most b + r*T. A general service curve bounds virtual delay; individual request delay additionally needs an ordering or scheduling guarantee. [2]

Packet boundary: fixed-rate, non-preemptive transmission and known packet length can tighten the FIFO response-time bound. That packet model is not automatically a storage-device model. [1]

Relations

related
Arrival curve, Token bucket, Tail latency, Strict service curve, Leftover service curve

Citations

  1. Mohammadpour, Stai and Le Boudec (2019), Improved Delay Bound for a Service Curve Element, sections II-IV (accessed 2026-11-20)
  2. Le Boudec and Thiran, Network Calculus, online edition 2022, sections 1.2-1.4 (accessed 2026-11-20)
Sources & caveats
  • Theory and worked examples, not a measured storage reservation. An average benchmark cannot establish a minimum-service guarantee.

JSON: /data/terms/rate-latency-service-curve.json

Published by Hesela team · Updated

Report a correction · Editorial standards · Reuse guidance

Cite this page
Hesela team. Rate-latency service curve. Hesela. 2026-11-20. https://hesela.dev/terms/rate-latency-service-curve/