RICA: Resource-Integrated Congestion Control for In-Network Aggregation
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.
Authors
- Wenfei Wu (ORCID: https://orcid.org/0000-0002-1357-3137)
- Jianglong Nie (ORCID: https://orcid.org/0009-0001-8492-7983)
- Yuan Yao (ORCID: https://orcid.org/0000-0002-2705-6245)
- Ziyu Wang (ORCID: https://orcid.org/0000-0002-4697-5183)
Institutions
- Peking University (CN)
Publication Details
- Journal
- Proceedings of the ACM on Networking
- Published
- 2026-11-19
- DOI
- https://doi.org/10.1145/3856973
- Primary Topic
- Network Traffic and Congestion Control
- Type
- article
- Field-Weighted Citation Impact
- 0.00