DMCA
Research Almaden • Austin • Beijing • Delhi • Haifa • T.J. Watson • Tokyo • ZurichFlow and Congestion Control for Datacenter Networks
BibTeX
@MISC{Gusat_researchalmaden,
author = {M. Gusat and C. Minkenberg and G. J. Paljak},
title = {Research Almaden • Austin • Beijing • Delhi • Haifa • T.J. Watson • Tokyo • ZurichFlow and Congestion Control for Datacenter Networks},
year = {}
}
Share
 |
 |
 |
 |
||
OpenURL
Abstract
Abstract: The limits of power dissipation and Moore's law are leading toward increasing parallelism and a shift of focus from CPUs to interconnection networks. This trend is also reflected in the rise of blade-based datacenters, which cluster server and storage units packaged as blades, with several networks. We begin with the trends and requirements of datacenter interconnection networks. Next, we show that lossless link-level flow control is a necessary feature of such networks, required for correctness and performance. However, such flow control schemes have a side-effect: saturation tree congestion, potentially causing catastrophic performance collapse. We argue that the ongoing trends toward increased efficiency, consolidation, and virtualization will escalate the likelihood of congestive collapse. This amplifies the need for congestion management with prevention and recovery mechanisms. Solutions established in best-effort networks (TCP/IP and ATM) are not directly suitable, mainly because they assume a lossy link layer. We advocate the need for research in flow and congestion control in lossless interconnects to meet the challenges of ubiquitous parallelism. 1. Pervasive Parallelism and Trends in Datacenters Advances in CPU and computer architecture in the last few years both in commercial-off-the-shelf products (e.g. multi-core chip multiprocessors) and in large-scale specialized systems (e.g. High-Performance Computing, HPC) renewed interest in parallelism, exploited at all levels: from VLSI to server packaging and from instruction- to widearea system-level grids. A fast growing example is scalable datacenters based on blade servers, described in [16]
Keyphrases
congestion control datacenter network tcp ip ongoing trend catastrophic performance collapse congestion management high-performance computing lossless link-level flow control increased efficiency datacenters advance last year computer architecture large-scale specialized system datacenter interconnection network blade server power dissipation commercial-off-the-shelf product best-effort network congestive collapse saturation tree congestion pervasive parallelism lossy link layer flow control scheme ubiquitous parallelism system-level grid multi-core chip multiprocessor lossless interconnects blade-based datacenters scalable datacenters recovery mechanism several network necessary feature storage unit
