Flexspander: augmenting expander networks in high-performance systems with optical bandwidth steering
Communication efficiency is one of the deciding factors in determining many of today’s high-performance computing (HPC) applications. Traditionally, HPC systems have been on static network topologies, making them inflexible to the variety of skewed traffic demands that may arise due to the spatial locality inherent in many applications. To handle traffic locality, researchers have proposed integrating optical circuit switches (OCSs) into the network architecture, which reconfigures the network topology to alter and dynamically adapt to the predicted traffic. In this paper, we present a novel reconfigurable network topology called Flexspander. Beyond offering a flexible interconnect, Flexspander also offers full flexibility in terms of construction and can be built with any arbitrary combination of commercial electrical packet switches and OCSs. We evaluate Flexspander performance through extensive simulations with multiple network traces, and our results show improved performance for Flexspander over currently proposed static and reconfigurable topologies in terms of the flow completion time.
- Sponsoring Organization:
- USDOE Advanced Research Projects Agency - Energy (ARPA-E)
- OSTI ID:
- 1601693
- Alternate ID(s):
- OSTI ID: 1803989
- Journal Information:
- Journal of Optical Communications and Networking, Journal Name: Journal of Optical Communications and Networking Journal Issue: 4 Vol. 12; ISSN JOCNBB; ISSN 1943-0620
- Publisher:
- Optical Society of AmericaCopyright Statement
- Country of Publication:
- United States
- Language:
- English
Ramanujan graphs
|
journal | September 1988 |
Fat-trees: Universal networks for hardware-efficient supercomputing
|
journal | October 1985 |
Three-dimensional Statistical Data Security Problems
|
journal | February 1994 |
Reconfigurable hybrid interconnection for static and dynamic scientific applications
|
conference | January 2007 |
A scalable, commodity data center network architecture
|
conference | January 2008 |
VL2: a scalable and flexible data center network
|
conference | January 2009 |
Understanding data center traffic characteristics
|
conference | January 2009 |
The nature of data center traffic: measurements & analysis
|
conference | January 2009 |
HyperX: topology, routing, and packaging of efficient large-scale networks
|
conference | January 2009 |
Network traffic characteristics of data centers in the wild
|
conference | January 2010 |
Integrating microsecond circuit switching into the data center
|
conference | August 2013 |
B4: experience with a globally-deployed software defined wan
|
conference | August 2013 |
pFabric: minimal near-optimal datacenter transport
|
conference | August 2013 |
Inside the Social Network's (Datacenter) Network
|
conference | August 2015 |
Xpander: Unveiling the Secrets of High-Performance Datacenters
|
conference | November 2015 |
ProjecToR: Agile Reconfigurable Data Center Interconnect
|
conference | August 2016 |
Xpander: Towards Optimal-Performance Datacenters
|
conference | January 2016 |
Bandwidth steering in HPC using silicon nanophotonics
|
conference | November 2019 |
Finding the K Shortest Loopless Paths in a Network
|
journal | July 1971 |
Entropy Waves, the Zig-Zag Graph Product, and New Constant-Degree Expanders
|
journal | January 2002 |
Similar Records
Machine-learning-aided cognitive reconfiguration for flexible-bandwidth HPC and data center networks [Invited]
Distributed deep learning training using silicon photonic switched architectures