ABSTRACT
With the increase of diversity in application needs and networks, existing congestion control algorithms (CCAs) do not accommodate this complicated reality. Previous classic CCAs are designed for a specific domain with fixed rules, failing to adapt to such diversities. Recently surged learning-based CCAs have great potential in adaptability and flexibility but are not practical due to unsatisfying performance on convergence, fairness, overhead and safety assurance. In this paper, we propose Libra, a unified congestion control framework, which empowers flexibility, adaptability, and practicality, by combining the wisdom of classic and reinforcement learning (RL)-based CCAs. Extensive evaluation of Libra's Linux kernel implementations on both live Internet and emulated networks shows performance improvement under dynamic networks (e.g., 1.2x throughput than Orca on average). At the same time, Libra can flexibly satisfy different application needs, reduce the running overhead by at most 0.92x and perform good fairness and convergence properties, well-fitting our theoretical analysis.
Supplemental Material
- Soheil Abbasloo, Yang Xu, and H. Jonathan Chao. 2019. C2TCP: A Flexible Cellular TCP to Meet Stringent Delay Requirements. IEEE J. Sel. Areas Commun. 37, 4 (2019), 918--932. Google ScholarCross Ref
- Soheil Abbasloo, Chen-Yu Yen, and H. Jonathan Chao. 2020. Classic Meets Modern: a Pragmatic Learning-Based Congestion Control for the Internet. In Proceedings of ACM SIGCOMM, Henning Schulzrinne and Vishal Misra (Eds.). ACM, 632--647. Google ScholarDigital Library
- Soheil Abbasloo, Chen-Yu Yen, and H. Jonathan Chao. 2021. Wanna Make Your TCP Scheme Great for Cellular Networks? Let Machines Do It for You! IEEE J. Sel. Areas Commun. 39, 1 (2021), 265--279. Google ScholarDigital Library
- Mohammad Alizadeh, Albert G. Greenberg, David A. Maltz, Jitendra Padhye, Parveen Patel, Balaji Prabhakar, Sudipta Sengupta, and Murari Sridharan. 2010. Data center TCP (DCTCP). In Proceedings of ACM SIGCOMM, Shivkumar Kalyanaraman, Venkata N. Padmanabhan, K. K. Ramakrishnan, Rajeev Shorey, and Geoffrey M. Voelker (Eds.). ACM, 63--74. Google ScholarDigital Library
- Venkat Arun and Hari Balakrishnan. 2018. Copa: Practical Delay-Based Congestion Control for the Internet. In Proceedings of USENIX NSDI, Sujata Banerjee and Srinivasan Seshan (Eds.). USENIX Association, 329--342. https://www.usenix.org/conference/nsdi18/presentation/arunGoogle ScholarDigital Library
- Lawrence S. Brakmo and Larry L. Peterson. 1995. TCP Vegas: End to End Congestion Avoidance on a Global Internet. IEEE J. Sel. Areas Commun. 13, 8 (1995), 1465--1480. Google ScholarDigital Library
- Lloyd Brown, Ganesh Ananthanarayanan, Ethan Katz-Bassett, Arvind Krishnamurthy, Sylvia Ratnasamy, Michael Schapira, and Scott Shenker. 2020. On the Future of Congestion Control for the Public Internet. In Proceedings of ACM HotNets, Ben Zhao, Heather Zheng, Harsha V. Madhyastha, and Venkat N. Padmanabhan (Eds.). ACM, 30--37. Google ScholarDigital Library
- Neal Cardwell, Yuchung Cheng, C. Stephen Gunn, Soheil Hassas Yeganeh, and Van Jacobson. 2017. BBR: congestion-based congestion control. Commun. ACM 60, 2 (2017), 58--66. Google ScholarDigital Library
- Kefan Chen, Danfeng Shan, Xiaohui Luo, Tong Zhang, Yajun Yang, and Fengyuan Ren. 2020. One Rein to Rule Them All: A Framework for Datacenter-to-User Congestion Control. In Proceedings of ACM APNet. ACM, 44--51. Google ScholarDigital Library
- A. V. Chernov. 2019. On Some Approaches to Find Nash Equilibrium in Concave Games. Autom. Remote. Control. 80, 5 (2019), 964--988. Google ScholarDigital Library
- Mo Dong, Qingxi Li, Doron Zarchy, Philip Brighten Godfrey, and Michael Schapira. 2015. PCC: Re-architecting Congestion Control for Consistent High Performance. In Proceedings of USENIX NSDI. USENIX Association, 395--408. https://www.usenix.org/conference/nsdi15/technical-sessions/presentation/dongGoogle Scholar
- Mo Dong, Tong Meng, Doron Zarchy, Engin Arslan, Yossi Gilad, Brighten Godfrey, and Michael Schapira. 2018. PCC Vivace: Online-Learning Congestion Control. In Proceedings of USENIX NSDI, Sujata Banerjee and Srinivasan Seshan (Eds.). USENIX Association, 343--356. https://www.usenix.org/conference/nsdi18/presentation/dongGoogle Scholar
- Salma Emara, Baochun Li, and Yanjiao Chen. 2020. Eagle: Refining Congestion Control by Learning from the Experts. In Proceedings of IEEE INFOCOM. IEEE, 676--685. Google ScholarDigital Library
- Eyal Even-Dar, Yishay Mansour, and Uri Nadav. 2009. On the convergence of regret minimization dynamics in concave games. In Proceedings of ACM STOC, Michael Mitzenmacher (Ed.). ACM, 523--532. Google ScholarDigital Library
- Sally Floyd and Thomas R. Henderson. 1999. The NewReno Modification to TCP's Fast Recovery Algorithm. RFC 2582 (1999), 1--12.Google Scholar
- Sangtae Ha, Injong Rhee, and Lisong Xu. 2008. CUBIC: a new TCP-friendly high-speed TCP variant. ACM SIGOPS Oper. Syst. Rev. 42, 5 (2008), 64--74. Google ScholarDigital Library
- Peter Henderson, Riashat Islam, Philip Bachman, Joelle Pineau, Doina Precup, and David Meger. 2018. Deep Reinforcement Learning That Matters. In Proceedings of AAAI, Sheila A. McIlraith and Kilian Q. Weinberger (Eds.). AAAI Press, 3207--3214. https://www.aaai.org/ocs/index.php/AAAI/AAAI18/paper/view/16669Google ScholarCross Ref
- Ashley Hill, Antonin Raffin, Maximilian Ernestus, Adam Gleave, Anssi Kanervisto, Rene Traore, Prafulla Dhariwal, Christopher Hesse, Oleg Klimov, Alex Nichol, Matthias Plappert, Alec Radford, John Schulman, Szymon Sidor, and Yuhuai Wu. 2018. Stable Baselines. https://github.com/hill-a/stable-baselines. (2018).Google Scholar
- Yannis E. Ioannidis and Eugene Wong. 1987. Query Optimization by Simulated Annealing. In Proceedings of the Association for Computing Machinery Special Interest Group on Management of Data, Umeshwar Dayal and Irving L. Traiger (Eds.). ACM Press, 9--22. Google ScholarDigital Library
- Nathan Jay, Noga H. Rotman, Brighten Godfrey, Michael Schapira, and Aviv Tamar. 2019. A Deep Reinforcement Learning Perspective on Internet Congestion Control. In Proceedings of PMLR ICML, Kamalika Chaudhuri and Ruslan Salakhutdinov (Eds.), Vol. 97. PMLR, 3050--3059. http://proceedings.mlr.press/v97/jay19a.htmlGoogle Scholar
- Yiming Kong, Hui Zang, and Xiaoli Ma. 2018. Improving TCP Congestion Control with Machine Intelligence. In Proceedings of the 2018 Workshop on Network Meets AI & ML, NetAI@SIGCOMM. ACM, 60--66. Google ScholarDigital Library
- Gautam Kumar, Nandita Dukkipati, Keon Jang, Hassan M. G. Wassel, Xian Wu, Behnam Montazeri, Yaogong Wang, Kevin Springborn, Christopher Alfeld, Michael Ryan, David Wetherall, and Amin Vahdat. 2020. Swift: Delay is Simple and Effective for Congestion Control in the Datacenter. In Proceedings of ACM SIGCOMM, Henning Schulzrinne and Vishal Misra (Eds.). ACM, 514--528. Google ScholarDigital Library
- Xu Li, Feilong Tang, Jiacheng Liu, Laurence T. Yang, Luoyi Fu, and Long Chen. 2021. AUTO: Adaptive Congestion Control Based on Multi-Objective Reinforcement Learning for the Satellite-Ground Integrated Network. In Proceedings of USENIX ATC, Irina Calciu and Geoff Kuenning (Eds.). USENIX Association, 611--624. https://www.usenix.org/conference/atc21/presentation/li-xuGoogle Scholar
- Yuxi Li. 2017. Deep Reinforcement Learning: An Overview. CoRR abs/1701.07274 (2017). arXiv:1701.07274 http://arxiv.org/abs/1701.07274Google Scholar
- Tong Meng, Neta Rozen Schiff, Philip Brighten Godfrey, and Michael Schapira. 2020. PCC Proteus: Scavenger Transport And Beyond. In Proceedings of ACM SIGCOMM, Henning Schulzrinne and Vishal Misra (Eds.). ACM, 615--631. Google ScholarDigital Library
- Ravi Netravali, Anirudh Sivaraman, Somak Das, Ameesh Goyal, Keith Winstein, James Mickens, and Hari Balakrishnan. 2015. Mahimahi: Accurate Record-and-Replay for HTTP. In Proceedings of USENIX ATC, Shan Lu and Erik Riedel (Eds.). USENIX Association, 417--429. https://www.usenix.org/conference/atc15/technical-session/presentation/netravaliGoogle Scholar
- Kathleen M. Nichols and Van Jacobson. 2012. Controlling queue delay. Commun. ACM 55, 7 (2012), 42--50. Google ScholarDigital Library
- Noga H. Rotman, Michael Schapira, and Aviv Tamar. 2020. Online Safety Assurance for Learning-Augmented Systems. In Proceedings of ACM HotNets, Ben Zhao, Heather Zheng, Harsha V. Madhyastha, and Venkat N. Padmanabhan (Eds.). ACM, 88--95. Google ScholarDigital Library
- Alessio Sacco, Matteo Flocco, Flavio Esposito, and Guido Marchetto. 2021. Owl: Congestion Control with Partially Invisible Networks via Reinforcement Learning. In Proceedings of IEEE INFOCOM. IEEE, 1--10. Google ScholarDigital Library
- John Schulman, Filip Wolski, Prafulla Dhariwal, Alec Radford, and Oleg Klimov. 2017. Proximal Policy Optimization Algorithms. CoRR abs/1707.06347 (2017). arXiv:1707.06347 http://arxiv.org/abs/1707.06347Google Scholar
- Anirudh Sivaraman, Keith Winstein, Pratiksha Thaker, and Hari Balakrishnan. 2014. An experimental study of the learnability of congestion control. In Proceedings of ACM SIGCOMM, Fabián E. Bustamante, Y. Charlie Hu, Arvind Krishnamurthy, and Sylvia Ratnasamy (Eds.). ACM, 479--490. Google ScholarDigital Library
- Kun Tan, Jingmin Song, Qian Zhang, and Murari Sridharan. 2006. A Compound TCP Approach for High-Speed and Long Distance Networks. In Proceedings of IEEE INFOCOM. IEEE. Google ScholarCross Ref
- Keith Winstein and Hari Balakrishnan. 2013. TCP ex machina: computergenerated congestion control. In Proceedings of ACM SIGCOMM, Dah Ming Chiu, Jia Wang, Paul Barford, and Srinivasan Seshan (Eds.). ACM, 123--134. Google ScholarDigital Library
- Keith Winstein, Anirudh Sivaraman, and Hari Balakrishnan. 2013. Stochastic Forecasts Achieve High Throughput and Low Delay over Cellular Networks. In Proceedings of USENIX NSDI, Nick Feamster and Jeffrey C. Mogul (Eds.). USENIX Association, 459--471. https://www.usenix.org/conference/nsdi13/technical-sessions/presentation/winsteinGoogle Scholar
- Yaxiong Xie, Fan Yi, and Kyle Jamieson. 2020. Pbe-CC: Congestion Control via Endpoint-Centric, Physical-Layer Bandwidth Measurements. In Proceedings of ACM SIGCOMM, Henning Schulzrinne and Vishal Misra (Eds.). ACM, 451--464. Google ScholarDigital Library
- Lisong Xu, Khaled Harfoush, and Injong Rhee. 2004. Binary Increase Congestion Control (BIC) for Fast Long-Distance Networks. In Proceedings of IEEE INFOCOM. IEEE, 2514--2524. Google ScholarCross Ref
- Zhiyuan Xu, Jian Tang, Chengxiang Yin, Yanzhi Wang, and Guoliang Xue. 2019. Experience-Driven Congestion Control: When Multi-Path TCP Meets Deep Reinforcement Learning. IEEE J. Sel. Areas Commun. 37, 6 (2019), 1325--1336. Google ScholarCross Ref
- Francis Y. Yan, Jestin Ma, Greg D. Hill, Deepti Raghavan, Riad S. Wahby, Philip Alexander Levis, and Keith Winstein. 2018. Pantheon: the training ground for Internet congestion-control research. In Proceedings of USENIX ATC, Haryadi S. Gunawi and Benjamin Reed (Eds.). USENIX Association, 731--743. https://www.usenix.org/conference/atc18/presentation/yan-francisGoogle Scholar
- Jiancheng Ye, Ka-Cheong Leung, Victor O. K. Li, and Steven H. Low. 2018. Combating Bufferbloat in Multi-Bottleneck Networks: Equilibrium, Stability, and Algorithms. In Proceedings of IEEE INFOCOM. IEEE, 648--656. Google ScholarDigital Library
- Yasir Zaki, Thomas Pötsch, Jay Chen, Lakshminarayanan Subramanian, and Carmelita Görg. 2015. Adaptive Congestion Control for Unpredictable Cellular Networks. In Proceedings of ACM SIGCOMM, Steve Uhlig, Olaf Maennel, Brad Karp, and Jitendra Padhye (Eds.). ACM, 509--522. Google ScholarDigital Library
- Doron Zarchy, Radhika Mittal, Michael Schapira, and Scott Shenker. 2017. An Axiomatic Approach to Congestion Control. In Proceedings of ACM HotNets, Sujata Banerjee, Brad Karp, and Michael Walfish (Eds.). ACM, 115--121. Google ScholarDigital Library
- Gaoxiong Zeng, Wei Bai, Ge Chen, Kai Chen, Dongsu Han, Yibo Zhu, and Lei Cui. 2019. Congestion Control for Cross-Datacenter Networks. In Proceedings of IEEE ICNP. IEEE, 1--12. Google ScholarCross Ref
Index Terms
- A unified congestion control framework for diverse application preferences and network conditions
Recommendations
On Explicit Congestion Notification for Stream Control Transmission Protocol in Lossy Networks
As a congestion avoidance mechanism, Explicit Congestion Notification (ECN) is designed to inform a data source to react to potential congestion early. Currently, the new transport protocol, Stream Control Transmission Protocol (SCTP), is not ECN-...
Performance analysis of loss-based high-speed TCP congestion control algorithms
CEA'09: Proceedings of the 3rd WSEAS international conference on Computer engineering and applicationsThe demands for fast transfer of large volumes of data, and the deployment of the network infrastructures to support the demand are ever increasing. However, the dominant transport protocol of today, TCP, does not meet this demand because it favors ...
Comparative analysis of TCP congestion control mechanisms
NISS '20: Proceedings of the 3rd International Conference on Networking, Information Systems & SecurityTCP (Transmission Control Protocol) is the most used transport protocol for wired and wireless networks. It provides many services (reliability, end to end delivery ...) to the applications running over the Internet, but to be able to manage traffics ...
Comments