Document Type

Technical Report

Department

Computer Science and Engineering

Publication Date

2007

Filename

wucse-2007-51.pdf

Technical Report Number

WUCSE-2007-51

Abstract

Modern high performance routers rely on sophisticated interconnection networks to meet ever increasing demands on capacity. Regulating the flow of packets through these interconnects is critical to providing good performance, particularly in the presence of extreme traffic patterns that result in sustained overload at output ports. Previous studies have used a combination of analysis and idealized simulations to show that coarse-grained scheduling of traffic flows can be effective in preventing congestion, while ensuring high utilization. In this paper, we study the performance of a coarse-grained scheduler in a real router with a scalable architecture similar to those found in high performance commercial systems. Our results are obtained by taking fine-grained measurements of an operating router that provide a detailed picture of how the scheduling algorithm behaves under a variety of conditions, giving a more complete and realistic understanding of the short time-scale dynamics than previous studies could provide. We also examine computation and communication overheads of our scheduler implementation to assess its resource usage and to provide the basis for an analysis of how the resource usage scales with system size.

Comments

Permanent URL: http://dx.doi.org/10.7936/K7QV3JSF

Download

Share

COinS