AIRP: Accelerating Multi-Tenant Distributed Learning With In-Network Resource Pooling
IEEE Transactions on Networking · 2026 · 34 · 4510-4525
DOI: 10.1109/ton.2026.3676228 ↗Kurz gefasst
Fachartikel in der Zeitschrift IEEE Transactions on Networking, 2026. Verfasst von Huifeng Xing und 9 weiteren Autor:innen. Die Arbeit nennt 67 Literaturangaben. Gefördert von: National Key Research and Development Program of China, National Natural Science Foundation of China, Natural Science Foundation of Shanghai.
Abstract (Originalsprache)
The increasing popularity of large models and datasets has highlighted the significance of distributed training networks. As gradient synchronization generates substantial traffic, in-network aggregation (INA) has emerged as a solution to offload aggregation onto the switch, alleviating network congestion and accelerating distributed training. However, the limited memory capacity of the INA switch becomes a potential bottleneck as computation shifts into the network, especially in multi-tenant scenarios. To address this bottleneck and enhance network throughput, we propose the Aggregation with In-network Resource Pooling (AIRP) framework. Unlike existing approaches that optimize individual switches in a localized manner, AIRP takes a holistic view and efficiently pools switch memory resources across the entire network, allocating them to multiple tenants. Evaluation using the ns-3 simulator and P4 testbed demonstrates that AIRP can accelerate the training of various models, including computer vision and language models. The experimental results show that AIRP outperforms existing INA approaches by up to 7 times in terms of network throughput in multi-tenant scenarios, while also achieving great flexibility and efficiency in deployment.
Förderung
- National Key Research and Development Program of China
- National Natural Science Foundation of China
- Natural Science Foundation of Shanghai
- Shanghai Magnolia Talent Plan Pujiang Project Funding
Zitieren
APA
Xing, H., Wang, H., Chen, Y., Hu, Y., Chen, Z., Liu, X., Li, Z., Shi, W., Liu, S., & Xu, Y. (2026). AIRP: Accelerating Multi-Tenant Distributed Learning With In-Network Resource Pooling. IEEE Transactions on Networking, 34, 4510-4525. https://doi.org/10.1109/ton.2026.3676228
Vancouver
Xing H, Wang H, Chen Y, Hu Y, Chen Z, Liu X, et al. AIRP: Accelerating Multi-Tenant Distributed Learning With In-Network Resource Pooling. IEEE Transactions on Networking. 2026;34:4510-4525. doi:10.1109/ton.2026.3676228
BibTeX
@article{Xing2026,
author = {Xing, Huifeng and Wang, Hao and Chen, Yang and Hu, Yinfan and Chen, Zixuan and Liu, Xuandong and Li, Zijian and Shi, Wanxin and Liu, Sen and Xu, Yang},
title = {AIRP: Accelerating Multi-Tenant Distributed Learning With In-Network Resource Pooling},
journal = {IEEE Transactions on Networking},
year = {2026},
volume = {34},
pages = {4510--4525},
doi = {10.1109/ton.2026.3676228}
}
EUVIMED