Work

On slowdown variance as a measure of fairness

Year: 2018

Type: article

Abstract: When considering fairness one must ask two fundamental questions. Firstly, what does it mean to be fair? And secondly, how does one measure that fairness? Different authors have offered different notions and metrics to address these questions. We provide arguments identifying where past metrics fall short, discuss how the underlying motivations differ, and offer our own metric to address these issues. That is, we propose using a system’s slowdown variance (SDV) as a measure for its fairness. Advantages of SDV are demonstrated via a suite of simulation experiments which compare a range of established policies under a range of service time distributions. These advantages include a decoupling of fairness from performance, an intuitive distinction between last come first serve and processor sharing , as well as recognition of starvation within shortest remaining processing time .

Source: Operations Research Perspectives

DOI: 10.1016/j.orp.2018.05.001

Authors: Vincent J. Maccio, Jenell Hogg, Douglas G. Down

Institution: McMaster University

Language: English

Topic: Advanced Queuing Theory Analysis

Subfield: Management Information Systems

Field: Business, Management and Accounting

Domain: Social Sciences

Open Access status: gold

Estimated APC paid: $1,790

Funder: Natural Sciences and Engineering Research Council of Canada