thesis-proposal

thesis-proposal - THESIS PROPOSAL A Theoretical Scheduling...

Info iconThis preview shows pages 1–6. Sign up to view the full content.

View Full Document Right Arrow Icon

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview: THESIS PROPOSAL A Theoretical Scheduling Toolbox Adam Wierman CMU-CS-05-??? School of Computer Science Carnegie Mellon University Pittsburgh, PA 15213 Abstract Scheduling policies are fundamental components of a majority of modern computer systems. However, de- spite a vast field of research analyzing the performance of different policies, the task of choosing a policy for a particular application is still difficult. This difficulty is a result of a disconnect between queueing re- searchers and system designers. Classical queueing research studies only individual scheduling policies and response time metrics; whereas practical issues force system designers to use hybrid policies that perform well across a variety of metrics. To bridge this divide, we propose a new style of scheduling research where large groups of policies are clas- sified with respect to a wide range of metrics. In particular, we propose to develop a theoretical scheduling toolbox consisting of a range of classifications, each of which isolates a different metric. The classifica- tions we propose can be divided into three types: classifications of efficiency metrics, fairness metrics, and robustness metrics. Classical queueing theory has focused primarily on measures of efficiency; thus, in developing the toolbox, we will need to develop novel metrics to measure the fairness and robustness of scheduling policies. In addition, because classical queueing research focuses on individual policies, we will need to develop novel analytic techniques in order to analyze the performance of large groups of policies. Keywords: scheduling; queueing; unfairness; efficiency; robustness; SRPT; LAS; FB; PS; SMART Thesis Committee Mor Harchol-Balter (Chair) Computer Science Department Carnegie Mellon University John Lafferty Computer Science Department Carnegie Mellon University Bruce Maggs Computer Science Department Carnegie Mellon University Alan Scheller-Wolf Tepper School of Business Carnegie Mellon University Ward Whitt Department of Industrial Engineering and Operations Research Columbia University i ii THESIS COMMITTEE Foreword This proposal is composed of three main sections. Chapter 1 contains a brief overview of the proposed thesis. This chapter is meant to condense the major motivation and impact of the work into a few short pages in order to give the reader an overview of what is ahead. Chapter 2 then presents a more thorough introduction to the work. The goal of this chapter is to convey the importance of scheduling in modern computer systems and to illustrate the impact of the scheduling toolbox we propose without delving into the details of the results. Finally, Chapter 3 provides a detailed summary of the proposed organization of the thesis, including a discussion of what results have been obtained so far and what results we hope to obtain over the next year and a half. Included at the close of the third chapter is a summary of the proposed timeline for the thesis....
View Full Document

This document was uploaded on 01/05/2012.

Page1 / 45

thesis-proposal - THESIS PROPOSAL A Theoretical Scheduling...

This preview shows document pages 1 - 6. Sign up to view the full document.

View Full Document Right Arrow Icon
Ask a homework question - tutors are online