The New Jersey Institute of Technology's
Electronic Theses & Dissertations Project

Title:	Preemptive efficient queueing for scheduling deadline-bound data center flows
Author:	GopalaKrishna, VinayKrishna
View Online:	njit-etd2017-067 (xi, 43 pages ~ 1.2 MB pdf)
Department:	Department of Electrical and Computer Engineering
Degree:	Master of Science
Program:	Internet Engineering
Document Type:	Thesis
Advisory Committee:	Rojas-Cessa, Roberto (Committee chair) Ansari, Nirwan (Committee member) Liu, Qing Gary (Committee member)
Date:	2017-05
Keywords:	Preemptive efficient queuing Data center flows
Availability:	Unrestricted
Abstract:	This thesis introduces a new deadline aware flow scheduling scheme called Preemptive Efficient Queuing (PEQ). Unlike other schemes which use policies like EDF, SJF or fair share, this scheme aims to procrastinate flows as much as they can afford to, while keeping with their deadlines. Thereby, PEQ prioritizes more urgent flows that have shorter time to procrastinate, or cushion time. Results of PEQ are compared with state-of-the-art schemes for the transport of data-center flows, such as Preemptive Distributed Quick (PDQ) and Deadline-Driven Delivery (D3). The results show that PEQ outperforms D3 and PDQ. We identify results of an optimal scheme and show that PDQ’s performance is close to that, yet not quite the same. Therefore, PEQ is a heuristic scheme with the equivalent complexity of PDQ but its performance is closer to the optimal solution than that of PDQ. The presented results show that when short flows have longer deadlines and long flows have stricter deadlines, the performance of PEQ, measured in terms of application throughput (i.e., the number of flows completed on time), is about twice that of PDQ.