Harmony: A Congestion-free Datacenter Architecture
Saksham Agarwal, Qizhe Cai, Rachit Agarwal, David B. Shmoys, Amin Vahdat
Abstract
Datacenter networks today provide best-effort deliverymessages may experience unpredictable queueing, delays, and drops due to switch buffer overflows within the network. Such weak guarantees reduce the set of assumptions that system designers can rely upon from the network, thus introducing inefficiency and complexity in host hardware and software.
We present Harmony, a datacenter network architecture that provides "congestion-free" message delivery guaranteeseach message, once transmitted by the sender, experiences bounded queueing at each switch in the network. Thus, by design, Harmony ensures that network delays are bounded in failure-free scenarios, and that congestion-related drops are eliminated. We establish, both theoretically and empirically, that Harmony provides these powerful properties with nearzero overheads compared to best-effort delivery networks: it incurs a tiny additive latency overhead that diminishes with message sizes, and achieves near-optimal network utilization.
Ask about this paper
Your agent reads all of it.
Lune indexed this paper to the last equation, along with the top-tier papers that cite it. Ask a question and the answer quotes them.
Your agent calls
Luneget_paper_fulltext
Free to start. No credit card required.
Terminal
Install the CLIlune papers fulltext ff3b2b99-8e4a-4266-8242-35be0d7adbe1Cited by top-tier papers8
- Fast, Scalable, and Accurate Rate Limiter for RDMA NICsZilong Wang, Xinchen Wan, Luyang Li, Yijun Sun et al.SIGCOMM 2024 · 17 citations
- Self-Clocked Round-Robin Packet SchedulingErfan Sharafzadeh, Raymond Matson, Jean Tourrilhes, Puneet Sharma et al.NSDI 2025 · 7 citations
- Building A CSFQ-Inspired Transport for Switched CXL Memory PoolingZerui Guo, Emily Shriver, Ming LiuNSDI 2026 · 2 citations
- SIRD: A Sender-Informed, Receiver-Driven Datacenter Transport ProtocolKonstantinos Prasopoulos, Ryan Kosta, Edouard Bugnion, Marios KogiasNSDI 2025
- OptiReduce: Resilient and Tail-Optimal AllReduce for Distributed Deep Learning in the CloudErtza Warraich, Omer Shabtai, Khalid Manaa, Shay Vargaftik et al.NSDI 2025
Builds on11
- Swift: Delay is Simple and Effective for Congestion Control in the DatacenterGautam Kumar, Nandita Dukkipati, Keon Jang, Hassan M. G. Wassel et al.SIGCOMM 2020 · 333 citations
- Understanding host network stack overheadsQizhe Cai, Shubham Chaudhary, Midhul Vuppalapati, Jaehyun Hwang et al.SIGCOMM 2021 · 150 citations
- Aeolus: A Building Block for Proactive Transport in DatacentersShuihai Hu, Wei Bai, Gaoxiong Zeng, Zilong Wang et al.SIGCOMM 2020 · 138 citations
- PowerTCP: Pushing the Performance Limits of Datacenter NetworksVamsi Addanki, Oliver Michel, Stefan SchmidNSDI 2022 · 116 citations
- Rearchitecting Linux Storage Stack for µs Latency and High ThroughputJaehyun Hwang, Midhul Vuppalapati, Simon Peter, Rachit AgarwalOSDI 2021 · 63 citations
Related papers
- Backpressure Flow ControlPrateesh Goyal, Preey Shah, Kevin Zhao, Georgios Nikolaidis et al.NSDI 2022
- Towards timeout-less transport in commodity datacenter networksHwijoon Lim, Wei Bai, Yibo Zhu, Youngmok Jung et al.EuroSys 2021 · 28 citations
- ACC: Addressing Performance Limitations in Datacenters with Atomic Congestion ControlZirui Wan, Jiao Zhang, Shuai Cheng, Ying Chen et al.INFOCOM 2025 · 3 citations
- Harmonia: Near-Linear Scalability for Replicated Storage with In-Network Conflict DetectionHang Zhu, Zhihao Bai, Jialin Li, Ellis Michael et al.VLDB 2020 · 58 citations
- Harmony: Saving Concurrent Transmissions from Harsh RF InterferenceXiaoyuan Ma, Peilin Zhang, Ye Liu, Carlo Alberto Boano et al.INFOCOM 2020 · 15 citations
