Liying Wang, Peking University; Jing Liu, ByteDance; Yuhan Zhou and Chengke Wang, Peking University; Mingming Lu, Qingyue Li, Song Geng, Linsen Wang, Kaida Hu, Haoyuan Huang, Shimao Tian, Ri Lu, and Mingfei Hao, ByteDance; Chenren Xu, Peking University; and Key Laboratory of High Confidence Software Technologies, Ministry of Education (PKU); Shu Shi, ByteDance
Live video streaming is a major source of today's Internet traffic, yet its delivery through CDNs incurs massive bandwidth costs from both CDN edge traffic (to users) and internal traffic (between CDN nodes). To understand these costs, we conduct a large-scale measurement study of ByteDance’s global in-production live CDN. It reveals two opportunities to reduce CDN bandwidth cost: (i) lowering average edge price by incorporating cost-effective best-effort infrastructure into the CDN edge, and (ii) improving CDN tree efficiency via finer-grained stream scheduling, accounting for detailed stream characteristics such as non-uniform popularity, heterogeneous stream formats, which amplify internal traffic on cache misses in our prior CDN. To exploit these opportunities into a practical system while addressing challenges in scheduling strategy management and user quality of experience, we design, implement, and deploy HCDN. It comprises (i) an augmented CDN edge with cost-effective best-effort nodes and multihomed nodes, (ii) the OpenTiga scheduling framework, which coordinates strategies through modular orchestration and proactive client-side Quality-of-Experience (QoE) assurance, and (iii) a set of redirection-based stream scheduling strategies. Over four years of incremental deployment across more than 500 edge clusters, HCDN reduces relative bandwidth cost by 36% while incurring modest overhead and preserving QoE. We further report the deployment experience of HCDN.
NSDI '26 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.

author = {Liying Wang and Jing Liu and Yuhan Zhou and Chengke Wang and Mingming Lu and Qingyue Li and Song Geng and Linsen Wang and Kaida Hu and Haoyuan Huang and Shimao Tian and Ri Lu and Mingfei Hao and Chenren Xu and Shu Shi},
title = {{HCDN}: Coordinated Stream Scheduling for {Cost-Effective} Live Video Delivery},
booktitle = {23rd USENIX Symposium on Networked Systems Design and Implementation (NSDI 26)},
year = {2026},
isbn = {978-1-939133-54-0},
address = {Renton, WA},
pages = {2603--2617},
url = {https://www.usenix.org/conference/nsdi26/presentation/wang-liying},
publisher = {USENIX Association},
month = may
}