RICA: Resource-Integrated Congestion Control for In-Network Aggregation
Proceedings of the ACM on Networking · 19.11.2026 · 4 (CoNEXT4) · 1-21
DOI: 10.1145/3856973 ↗Kurz gefasst
Fachartikel in der Zeitschrift Proceedings of the ACM on Networking, online erschienen am 19.11.2026. Verfasst von Ziyu Wang und 3 weiteren Autor:innen. Die Arbeit nennt 38 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Beijing Municipal Natural Science Foundation.
Abstract (Originalsprache)
In-network aggregation (INA) reduces parameter-server traffic in distributed DNN training, but best-effort INA makes congestion control harder than ordinary packet forwarding. When a packet collides with an occupied switch aggregator, it falls back to end-host aggregation and becomes additional traffic on the parameter-server-bound link. Existing INA transports either react to this mixed load with coarse link-level control or manage aggregator collisions separately, missing the coupling between congestion window, collision probability, and fallback traffic. This paper presents RICA, a resource-integrated congestion control framework for best-effort INA. RICA models the bottleneck load as the sum of aggregated traffic and collision-induced fallback traffic, and derives an adaptive backoff factor from this model. RICA also adds lightweight in-band straggler telemetry to existing ACKs, allowing fast workers to yield aggregator occupancy when a persistent straggler is detected. Packet-level simulations show that RICA reduces average training iteration time by up to 8.6%--12.0% compared with the state-of-art INA system. Our code is available at https://github.com/wzy240/rica.
Förderung
- Beijing Municipal Natural Science Foundation
Zitieren
APA
Wang, Z., Nie, J., Yao, Y., & Wu, W. (2026). RICA: Resource-Integrated Congestion Control for In-Network Aggregation. Proceedings of the ACM on Networking, 4(CoNEXT4), 1-21. https://doi.org/10.1145/3856973
Vancouver
Wang Z, Nie J, Yao Y, Wu W. RICA: Resource-Integrated Congestion Control for In-Network Aggregation. Proceedings of the ACM on Networking. 2026;4(CoNEXT4):1-21. doi:10.1145/3856973
BibTeX
@article{Wang2026,
author = {Wang, Ziyu and Nie, Jianglong and Yao, Yuan and Wu, Wenfei},
title = {RICA: Resource-Integrated Congestion Control for In-Network Aggregation},
journal = {Proceedings of the ACM on Networking},
year = {2026},
volume = {4},
number = {CoNEXT4},
pages = {1--21},
doi = {10.1145/3856973}
}
EUVIMED