R2D2: bufferless, switchless data center networks using commodity ethernet hardware

Authors:
Matthew P. Grosvenor;Malte Schwarzkopf;Andrew W. Moore
Affiliations:
University of Cambridge Computer Laboratory, Cambridge, United Kingdom;University of Cambridge Computer Laboratory, Cambridge, United Kingdom;University of Cambridge Computer Laboratory, Cambridge, United Kingdom
Venue:
Proceedings of the ACM SIGCOMM 2013 conference on SIGCOMM
Year:
2013

Citing 9
Cited 0

BCube: a high performance, server-centric network architecture for modular data centers

Proceedings of the ACM SIGCOMM 2009 conference on Data communication
Understanding TCP incast throughput collapse in datacenter networks

Proceedings of the 1st ACM workshop on Research on enterprise networking
Data center TCP (DCTCP)

Proceedings of the ACM SIGCOMM 2010 conference
Better never than late: meeting deadlines in datacenter networks

Proceedings of the ACM SIGCOMM 2011 conference
Less is more: trading a little bandwidth for ultra-low latency in the data center

NSDI'12 Proceedings of the 9th USENIX conference on Networked Systems Design and Implementation
Deadline-aware datacenter tcp (D2TCP)

Proceedings of the ACM SIGCOMM 2012 conference on Applications, technologies, architectures, and protocols for computer communication
Finishing flows quickly with preemptive scheduling

Proceedings of the ACM SIGCOMM 2012 conference on Applications, technologies, architectures, and protocols for computer communication
Deconstructing datacenter packet transport

Proceedings of the 11th ACM Workshop on Hot Topics in Networks
The tail at scale

Communications of the ACM

Quantified Score

Hi-index	0.00

Visualization

Abstract

Modern data centers commonly run distributed applications that require low-latency communication, and whose performance is critical to service revenue. If as little as one machine in 10,000 is a latency outlier, around 18% of requests will experience high latency. The sacrifice of latency determinism for bandwidth, however, is not an inevitable one. In our R2D2 architecture, we conceptually split the data centre network into an unbuffered, unswitched low-latency network (LLNet) and a deeply buffered bandwidth centric network (BBNet). Through explicitly scheduling network multiplexing in software, our prototype implementation achieves 99.995% and 99.999% messaging latencies of 35us and 75us respectively for 1514-byte packets on a fully loaded network. Furthermore, we show that it is possible to merge the conceptually separate LLNet and BBNet networks onto the same physical infrastructure using commodity switched Ethernet hardware.