DOE PAGES title logo U.S. Department of Energy
Office of Scientific and Technical Information

Title: Toward higher-radix switches with co-packaged optics for improved network locality in data center and HPC networks [Invited]

Journal Article · · Journal of Optical Communications and Networking

In this work, we study the network locality improvements that can be achieved by using co-packaged optics in data center and high-performance computing (HPC) networks. The increased escape bandwidth offered by co-packaged optics can enable switches with speeds of 51.2 Tb/s and beyond. From a network architecture perspective, the key advantages of introducing co-packaged optics at the switch points include the implementation of large-scale topologies of >12,000 end points with 4× higher bisection bandwidth and the reduction of the required number of switches by >40% compared with state-of-the-art approaches. From a network operation perspective, improved network locality and faster operation can be achieved since the higher-radix switches can mitigate the impact of network contention. Placing applications under fewer leaf switches reduces the number of packets that cross the spine switches in a leaf-spine topology. The proposed scheme is evaluated via discrete-event simulations: we initially evaluate the network locality properties of the system by using virtual-machine traces from a production data center, and we subsequently quantify the performance improvements by simulating an all-to-all pattern for a variety of message sizes over a number of nodes. The results suggest that co-packaged optics form a promising solution for keeping up with bandwidth scaling in future networks. The virtual-machine analysis shows that large-scale applications can be placed under up to 50% fewer first-level switches, while the network analysis shows speedups of up to 7.1, which translates to execution time reductions of up to 26% and 42.7% for applications with communication ratios of 0.3 and 0.5, respectively.

Research Organization:
IBM, Yorktown Heights, NY (United States). Thomas J. Watson Research Center
Sponsoring Organization:
USDOE Advanced Research Projects Agency - Energy (ARPA-E)
Grant/Contract Number:
AR0000846
OSTI ID:
1855243
Journal Information:
Journal of Optical Communications and Networking, Journal Name: Journal of Optical Communications and Networking Journal Issue: 6 Vol. 14; ISSN 1943-0620
Publisher:
IEEE - Optical Society of America (OSA)Copyright Statement
Country of Publication:
United States
Language:
English

References (22)

How to slice and dice your switch capacity conference January 2019
Co‐packaged datacenter optics: Opportunities and challenges journal March 2021
On the optimum switch radix in fat tree networks conference July 2011
1.6 Tbps Silicon Photonics Integrated Circuit and 800 Gbps Photonic Engine for Switch Co-Packaging Demonstration journal February 2021
TeraPHY: A Chiplet Technology for Low-Power, High-Bandwidth In-Package Optical I/O journal March 2020
Communication Scheduling Optimization for Distributed Deep Learning Systems conference December 2018
Input Versus Output Queueing on a Space-Division Packet Switch journal December 1987
Co-packaged optics for HPC and data center networks conference March 2021
Bandwidth-optimal all-to-all exchanges in fat tree networks
  • Prisacari, Bogdan; Rodriguez, German; Minkenberg, Cyriel
  • Proceedings of the 27th international ACM conference on International conference on supercomputing - ICS '13 https://doi.org/10.1145/2464996.2465434
conference January 2013
RotorNet: A Scalable, Low-complexity, Optical Datacenter Network
  • Mellette, William M.; McGuinness, Rob; Roy, Arjun
  • SIGCOMM '17: ACM SIGCOMM 2017 Conference, Proceedings of the Conference of the ACM Special Interest Group on Data Communication https://doi.org/10.1145/3098822.3098838
conference August 2017
Bandwidth steering in HPC using silicon nanophotonics
  • Michelogiannakis, George; Shen, Yiwen; Teh, Min Yee
  • SC '19: The International Conference for High Performance Computing, Networking, Storage, and Analysis, Proceedings of the International Conference for High Performance Computing, Networking, Storage and Analysis https://doi.org/10.1145/3295500.3356145
conference November 2019
Electronic packaging of the IBM z13 processor drawer journal July 2015
IBM POWER9 package technology and design journal July 2018
High-Port and Low-Latency Optical Switches for Disaggregated Data Centers: The Hipoλaos Switch Architecture [Invited] journal January 2018
Reimagining Datacenter Topologies With Integrated Silicon Photonics journal January 2018
Toward lower-diameter large-scale HPC and data center networks with co-packaged optics journal November 2020
OPSquare: A Flat DCN Architecture Based on Flow-Controlled Optical Packet Switches journal January 2017
Network Architecture in the Era of Integrated Optics conference January 2018
CloudSim Plus: A cloud computing simulation framework pursuing software engineering principles for improved modularity, extensibility and correctness conference May 2017
High Speed VCSELs and Co-Packaging for Short Reach Communication within Cloud and High Performance Computing conference November 2019
Big Data and Deep Learning Platform for Terabyte-Scale Renewable Datasets conference June 2018
Towards Massively Parallel Simulations of Massively Parallel High-Performance Computing Systems conference January 2012