TY - GEN
T1 - Harmony
T2 - 21st USENIX Symposium on Networked Systems Design and Implementation, NSDI 2024
AU - Agarwal, Saksham
AU - Cai, Qizhe
AU - Agarwal, Rachit
AU - Shmoys, David
AU - Vahdat, Amin
N1 - Publisher Copyright:
© 2024 Proceedings of the 21st USENIX Symposium on Networked Systems Design and Implementation, NSDI 2024. All rights reserved.
PY - 2024
Y1 - 2024
N2 - Datacenter networks today provide best-effort delivery—messages 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 guarantees—each 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 near-zero 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.
AB - Datacenter networks today provide best-effort delivery—messages 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 guarantees—each 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 near-zero 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.
UR - http://www.scopus.com/inward/record.url?scp=85194144149&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85194144149&partnerID=8YFLogxK
M3 - Conference contribution
AN - SCOPUS:85194144149
T3 - Proceedings of the 21st USENIX Symposium on Networked Systems Design and Implementation, NSDI 2024
SP - 329
EP - 343
BT - Proceedings of the 21st USENIX Symposium on Networked Systems Design and Implementation, NSDI 2024
PB - USENIX Association
Y2 - 16 April 2024 through 18 April 2024
ER -