A Transient Queueing Analysis under Time-varying Arrival and Service Rates for Enabling Low-Latency Services

TL;DR

This paper introduces a new analytical framework for modeling transient queue behavior with time-varying arrival and service rates, crucial for designing low-latency Internet services, validated through simulations.

Contribution

The paper develops a novel analytical method for transient queue analysis under dynamic conditions, enabling precise predictions of queue metrics in short-lived sessions.

Findings

Framework accurately predicts queue length distributions over time.

Validation confirms high accuracy against simulation results.

Demonstrates practical utility in low-latency service design.

Abstract

Understanding the detailed queueing behavior of a networking session is critical in enabling low-latency services over the Internet. Especially when the packet arrival and service rates at the queue of a link vary over time and moreover when the session is short-lived, analyzing the corresponding queue behavior as a function of time, which involves a transient analysis, becomes extremely challenging. In this paper, we propose and develop a new analytical framework that anatomizes the transient queue behavior under time-varying arrival and service rates even under unstable conditions. Our framework is capable of answering key questions in designing low-latency services such as the time-dependent probability distribution of the queue length; the instantaneous or time-averaged violation probability that the queue length exceeds a certain threshold; and the fraction of time during an interval $[0, t]$ at which the queue length exceeds a certain threshold. We validate our framework by comparing its prediction results over time with the statistical simulation results and confirm that our analysis is accurate enough. Our extensive demonstrations on the efficacy of the analytical framework in designing low-latency services reveal that its prediction ability for the transient queue behavior in diverse time-varying packet arrival and service patterns can be of a high practical value.