A switch-based approach to throughput collapse and starvation in data centers

Alexander Shpiner; Isaac Keslassy; Gabi Bracha; Eyal Dagan; Ofer Iny; Eyal Soha

doi:10.1016/j.comnet.2012.06.001

A switch-based approach to throughput collapse and starvation in data centers

Alexander Shpiner, Isaac Keslassy, Gabi Bracha, Eyal Dagan, Ofer Iny, Eyal Soha

Electrical and Computer Engineering

Technion - Israel Institute of Technology

Research output: Contribution to journal › Article › peer-review

Abstract

Data center switches need to satisfy stringent low-delay and high-capacity requirements. To do so, they rely on small switch buffers. However, in case of congestion, data center switches may suffer from throughput collapse for short TCP flows as well as temporary starvation for long TCP flows. In this paper, we introduce a lightweight hash-based algorithm called HCF (Hashed Credits Fair) to solve these problems at the switch level while being transparent to the end users. We show that it can be readily implemented in data center switches with O(1) complexity and negligible overhead. We illustrate using simulations how HCF mitigates the throughput collapse of short flows. We also show how HCF reduces unfairness and starvation for long-lived TCP flows as well as for short TCP flows, yet maximizes the utilization on the congested link. Last, HCF also prevents packet reordering.

Original language	English
Pages (from-to)	3333-3346
Number of pages	14
Journal	Computer Networks
Volume	56
Issue number	14
DOIs	https://doi.org/10.1016/j.comnet.2012.06.001
State	Published - 28 Sep 2012

Keywords

Data centers
Switching
TCP

All Science Journal Classification (ASJC) codes

Computer Networks and Communications

Access to Document

10.1016/j.comnet.2012.06.001

Cite this

@article{e5a97fca60a5457ca54f049a2f5db4e6,

title = "A switch-based approach to throughput collapse and starvation in data centers",

abstract = "Data center switches need to satisfy stringent low-delay and high-capacity requirements. To do so, they rely on small switch buffers. However, in case of congestion, data center switches may suffer from throughput collapse for short TCP flows as well as temporary starvation for long TCP flows. In this paper, we introduce a lightweight hash-based algorithm called HCF (Hashed Credits Fair) to solve these problems at the switch level while being transparent to the end users. We show that it can be readily implemented in data center switches with O(1) complexity and negligible overhead. We illustrate using simulations how HCF mitigates the throughput collapse of short flows. We also show how HCF reduces unfairness and starvation for long-lived TCP flows as well as for short TCP flows, yet maximizes the utilization on the congested link. Last, HCF also prevents packet reordering.",

keywords = "Data centers, Switching, TCP",

author = "Alexander Shpiner and Isaac Keslassy and Gabi Bracha and Eyal Dagan and Ofer Iny and Eyal Soha",

note = "Funding Information: Isaac Keslassy received his M.S. and Ph.D. degrees in Electrical Engineering from Stanford University, Stanford, CA, in 2000 and 2004, respectively. He is currently an associate professor in the Electrical Engineering department of the Technion, Haifa, Israel. His recent research interests include the design and analysis of high-performance routers and on-chip networks. The recipient of the European Research Council Starting Grant, the Yigal Alon Fellowship and the ATS-WD Career Development Chair, he is an associate editor for the IEEE/ACM Transactions on Networking. Funding Information: The conference version of this paper received the best paper award at IWQoS{\textquoteright}10, Beijing, China, June 2010. This work was partly supported by European Research Council Starting Grant No. 210389 and by the Hasso Plattner Institute Research School.",

year = "2012",

month = sep,

day = "28",

doi = "10.1016/j.comnet.2012.06.001",

language = "الإنجليزيّة",

volume = "56",

pages = "3333--3346",

journal = "Computer Networks",

issn = "1389-1286",

publisher = "Elsevier B.V.",

number = "14",

}

TY - JOUR

T1 - A switch-based approach to throughput collapse and starvation in data centers

AU - Shpiner, Alexander

AU - Keslassy, Isaac

AU - Bracha, Gabi

AU - Dagan, Eyal

AU - Iny, Ofer

AU - Soha, Eyal

N1 - Funding Information: Isaac Keslassy received his M.S. and Ph.D. degrees in Electrical Engineering from Stanford University, Stanford, CA, in 2000 and 2004, respectively. He is currently an associate professor in the Electrical Engineering department of the Technion, Haifa, Israel. His recent research interests include the design and analysis of high-performance routers and on-chip networks. The recipient of the European Research Council Starting Grant, the Yigal Alon Fellowship and the ATS-WD Career Development Chair, he is an associate editor for the IEEE/ACM Transactions on Networking. Funding Information: The conference version of this paper received the best paper award at IWQoS’10, Beijing, China, June 2010. This work was partly supported by European Research Council Starting Grant No. 210389 and by the Hasso Plattner Institute Research School.

PY - 2012/9/28

Y1 - 2012/9/28

N2 - Data center switches need to satisfy stringent low-delay and high-capacity requirements. To do so, they rely on small switch buffers. However, in case of congestion, data center switches may suffer from throughput collapse for short TCP flows as well as temporary starvation for long TCP flows. In this paper, we introduce a lightweight hash-based algorithm called HCF (Hashed Credits Fair) to solve these problems at the switch level while being transparent to the end users. We show that it can be readily implemented in data center switches with O(1) complexity and negligible overhead. We illustrate using simulations how HCF mitigates the throughput collapse of short flows. We also show how HCF reduces unfairness and starvation for long-lived TCP flows as well as for short TCP flows, yet maximizes the utilization on the congested link. Last, HCF also prevents packet reordering.

AB - Data center switches need to satisfy stringent low-delay and high-capacity requirements. To do so, they rely on small switch buffers. However, in case of congestion, data center switches may suffer from throughput collapse for short TCP flows as well as temporary starvation for long TCP flows. In this paper, we introduce a lightweight hash-based algorithm called HCF (Hashed Credits Fair) to solve these problems at the switch level while being transparent to the end users. We show that it can be readily implemented in data center switches with O(1) complexity and negligible overhead. We illustrate using simulations how HCF mitigates the throughput collapse of short flows. We also show how HCF reduces unfairness and starvation for long-lived TCP flows as well as for short TCP flows, yet maximizes the utilization on the congested link. Last, HCF also prevents packet reordering.

KW - Data centers

KW - Switching

KW - TCP

UR - http://www.scopus.com/inward/record.url?scp=84865730851&partnerID=8YFLogxK

U2 - 10.1016/j.comnet.2012.06.001

DO - 10.1016/j.comnet.2012.06.001

M3 - مقالة

SN - 1389-1286

VL - 56

SP - 3333

EP - 3346

JO - Computer Networks

JF - Computer Networks

IS - 14

ER -

A switch-based approach to throughput collapse and starvation in data centers

Abstract

Keywords

All Science Journal Classification (ASJC) codes

Access to Document

Other files and links

Fingerprint

Cite this