Abstract
High-performance computing (HPC) and data centers are showing a trend of merging into high-performance data centers (HPDC). HPDC is committed to providing extremely low latency for HPC or data center workloads. In addition to adopting a low-diameter network topology, HPDC also requires a more advanced congestion control mechanism. This paper implements the state-of-the-art congestion control method in the data centers on the Dragonfly topology and proposes Bowshot, a fast and accurate congestion control method for low-latency HPDC. Bowshot uses fine-grained feedback to accurately describe the network state. It uses switch feedback, ACK-padding, and ACK-first to reduce feedback delay. Bowshot uses switch calculation to reduce the overhead of congestion control. As the large-scale evaluation shows, Bowshot reduced the average flow completion time (FCT) by 33% and the 99th percentile FCT by 45% compared to the state-of-the-art work. Bowshot reduces the feedback delay by 89%. In addition, Bowshot maintains higher throughput and a near-zero queue length.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Similar content being viewed by others
References
INET (2021). https://inet.omnetpp.org/
OMNeT++ (2021). https://omnetpp.org
Alizadeh, M., et al.: Data center TCP (DCTCP). In: Proceedings of the ACM SIGCOMM (2010)
Bailey, D., Harris, T., Saphir, W., Van Der Wijngaart, R., Woo, A., Yarrow, M.: The NAS parallel benchmarks 2.0. Technical report, Technical Report NAS-95-020, NASA Ames Research Center (1995)
Chunduri, S., et al.: GPCNeT: designing a benchmark suite for inducing and measuring contention in HPC networks. In: Proceedings of the SC (2019)
De Sensi, D., Di Girolamo, S., Hoefler, T.: Mitigating network noise on dragonfly networks through application-aware routing. In: Proceedings of the SC (2019)
De Sensi, D., Di Girolamo, S., McMahon, K., Roweth, D., Hoefler, T.: An in-depth analysis of the slingshot interconnect (September 2020)
Dean, J., Barroso, L.A.: The tail at scale. Commun. ACM 56(2), 74–80 (2013)
Faanes, G., et al.: Cray cascade: a scalable HPC system based on a dragonfly network. In: Proceedings of the SC (2012)
Floyd, S.: TCP and explicit congestion notification. ACM SIGCOMM Comput. Commun. Rev. 24(5), 8–23 (1994)
Jiang, N., Becker, D.U., Michelogiannakis, G., Dally, W.J.: Network congestion avoidance through speculative reservation. In: Proceedings of the ISCA (2012)
Jiang, N., Dennison, L., Dally, W.J.: Network endpoint congestion control for fine-grained communication. In: Proceedings of the SC (2015)
Kathareios, G., Minkenberg, C., Prisacari, B., Rodriguez, G., Hoefler, T.: Cost-effective diameter-two topologies: analysis and evaluation. In: Proceedings of the SC (2015)
Kim, C., Sivaraman, A., Katta, N., Bas, A., Dixit, A., Wobker, L.J.: In-band network telemetry via programmable dataplanes. In: Proceedings of the ACM SIGCOMM (2015)
Kim, J., Dally, W.J., Scott, S., Abts, D.: Technology-driven, highly-scalable dragonfly topology. In: Proceedings of the ISCA (2008)
Lee, C., Park, C., Jang, K., Moon, S., Han, D.: Accurate latency-based congestion feedback for datacenters. In: Proceedings of the USENIX ATC (2015)
Leiserson, C.E.: Fat-trees: universal networks for hardware-efficient supercomputing. IEEE Trans. Comput. 100(10), 892–901 (1985)
Li, Y., et al.: HPCC: high precision congestion control. In: Proceedings of the ACM SIGCOMM (2019)
Misra, P.A., Borge, M.F., Goiri, Í., Lebeck, A.R., Zwaenepoel, W., Bianchini, R.: Managing tail latency in datacenter-scale file systems under production constraints. In: Proceedings of the EuroSys (2019)
Mittal, R., et al.: Timely: RTT-based congestion control for the datacenter. In: Proceedings of the ACM SIGCOMM (2015)
Roy, A., Zeng, H., Bagga, J., Porter, G., Snoeren, A.C.: Inside the social network’s (datacenter) network. In: Proceedings of the ACM SIGCOMM (2015)
Santos, J.R., Turner, Y., Janakiraman, G.: End-to-end congestion control for infiniband. In: Proceedings of the IEEE INFOCOM (2003)
Thaler, P.: IEEE 802.1Qau congestion notification (2006)
Wu, K., Dong, D., Li, C., Huang, S., Dai, Y.: Network congestion avoidance through packet-chaining reservation. In: Proceedings of the ICPP (2019)
Zhou, R., Dong, D., Huang, S., Bai, Y.: FastTune: timely and precise congestion control in data center network. In: Proceedings of the IEEE ISPA (2021)
Zhou, R., Yuan, G., Dong, D., Huang, S.: APCC: agile and precise congestion control in datacenters. In: Proceedings of the IEEE ISPA (2020)
Zhu, Y., et al.: Congestion control for large-scale RDMA deployments. In: Proceedings of the ACM SIGCOMM (2015)
Acknowledgements
We would like to thank the anonymous reviewers for their insightful comments. The work was supported by the National Key R&D Program of China under Grant No. 2018YFB0204300, Excellent Youth Foundation of Hunan Province (Dezun Dong) and National Postdoctoral Program for Innovative Talents Grant No. BX20190091.
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2022 IFIP International Federation for Information Processing
About this paper
Cite this paper
Zhou, R., Dong, D., Huang, S., Zhou, Z., Bai, Y. (2022). Taming Congestion and Latency in Low-Diameter High-Performance Datacenters. In: Cérin, C., Qian, D., Gaudiot, JL., Tan, G., Zuckerman, S. (eds) Network and Parallel Computing. NPC 2021. Lecture Notes in Computer Science(), vol 13152. Springer, Cham. https://doi.org/10.1007/978-3-030-93571-9_18
Download citation
DOI: https://doi.org/10.1007/978-3-030-93571-9_18
Published:
Publisher Name: Springer, Cham
Print ISBN: 978-3-030-93570-2
Online ISBN: 978-3-030-93571-9
eBook Packages: Computer ScienceComputer Science (R0)