Harmony: A Congestion-free Datacenter Architecture

Authors: 

Saksham Agarwal, Qizhe Cai, Rachit Agarwal, and David Shmoys, Cornell University; Amin Vahdat, Google

Abstract: 

Datacenter networks today provide best-effort delivery—messages may observe 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 powerful "congestion-free" message delivery guarantees—each message, once transmitted by the sender, observes bounded queueing at each switch in the network. Thus, network delays are bounded in failure-free scenarios, and congestion-related drops are completely eliminated. We establish, both theoretically and empirically, that Harmony provides such powerful guarantees with near-zero overheads compared to best-effort delivery networks: it incurs a tiny additive latency overhead that diminishes with message sizes, while achieving near-optimal network utilization.

NSDI '24 Open Access Sponsored by
King Abdullah University of Science and Technology (KAUST)

Open Access Media

USENIX is committed to Open Access to the research presented at our events. Papers and proceedings are freely available to everyone once the event begins. Any video, audio, and/or slides that are posted after the event are also free and open to everyone. Support USENIX and our commitment to Open Access.

BibTeX
@inproceedings {295507,
author = {Saksham Agarwal and Qizhe Cai and Rachit Agarwal and David Shmoys and Amin Vahdat},
title = {Harmony: A Congestion-free Datacenter Architecture},
booktitle = {21st USENIX Symposium on Networked Systems Design and Implementation (NSDI 24)},
year = {2024},
isbn = {978-1-939133-39-7},
address = {Santa Clara, CA},
pages = {329--343},
url = {https://www.usenix.org/conference/nsdi24/presentation/agarwal-saksham},
publisher = {USENIX Association},
month = apr
}

Presentation Video